A Severity Analysis of Accidents of Delivery Riders Based on Partial Proportional Odds Model
-
摘要: 针对外卖骑手事故严重度的致因分析,现有研究以骑手、行为、时间、空间、道路、环境、事故特征的部分因素为主,未从这7个方面同时量化不同因素对外卖事故严重度的影响差异性,特别是未考虑外卖类型、入口数、交叉口角度、舒适指数等因素。此外,当自变量中同时具有无序与有序的二分类或多分类变量时,既有模型将受到平行线假设限制,无法灵活地允许部分变量遵守而其余变量违背此假设。选取西安市1 473起外卖骑手交通事故,剖析事故严重度分布特性与时空分布特征,从以上7个方面综合选取25个潜在影响因素,构建部分优势比模型来揭示不同因素对外卖事故中骑手受伤严重度的影响显著性与平行线假设违背状况,并结合其边际效应分析来量化各显著因素间与因素内对此事故严重度的影响差异性。结果表明:外卖骑手事故严重度时间分布存在“双高峰”现象,城区事故密度大于郊区,路段上骑手轻伤占比(35.57%)大于交叉口处(31.76%);部分优势比模型效果优于有序Logit模型与广义有序Logit模型,外卖类型、季节、位置、入口数、交叉口角度、路面、天气与舒适度均遵守平行线假设;不同显著因素对外卖骑手事故严重度存在显著影响差异,闯红灯、逆行与超速等违章行为对外卖骑手受伤严重度影响最大,其边际效应绝对值最大值超过51%,而未曾被探析的入口数、交叉口角度、自行车专用道、外卖类型、舒适度等显著因素对外卖骑手受伤严重度也存在较大影响(8%~37%)。Abstract: The existing studies of causal analysis on accidents severity of delivery riders mainly focus on partial fac-tors such as rider, behavior, time, space, road, environment, and accident characteristics. The differences are not quantified in the impacts of different factors on the severity of accidents from the seven aspects above, without con-sidering factors such as takeaway type, number of entrances, angle of intersection, comfort index, and other factors. In addition, when there are both unordered and ordered binary or multi-categorical variables in the independent vari-ables, the established models are limited by the parallel-lines (PL) assumption, and fail to own the flexibility of al-lowing some variables to comply, while others violate this assumption. A total of 1 473 accidents related with deliv-ery riders in Xi'an are selected to analyze the severity distribution and the spatiotemporal distribution. A total of 25 potential influencing factors are selected from the seven aspects. A partial proportional odds (PPO) model is developed to clarify the significant influence of various factors on the rider injury severities in delivery crashes and the vi-olation status of the PL assumption. The corresponding marginal effects are carried out to quantify the differences between and within the contributing factors. The results show that there is a"double peak"phenomenon in the tem-poral distribution of the delivery-involved crashes, and the crash density in urban areas is higher than in suburban ar-eas. The proportion of minor injuries to riders on road sections (35.57%) is higher than at intersections (31.76%). The PPO model performs better than the ordered Logit model and the generalized ordered Logit model. The deliv-ery type, season, location, number of entrances, intersection angle, road surface, weather, and comfort index all fol-low the PL assumption. There are significant differences in the crash severity among different significant factors. Traffic violations such as running red lights, going in the wrong direction, and speeding have the greatest impacts on the rider crash severity, with the maximum absolute value of the marginal effects exceeding 51%. However, some unexplored factors such as number of entrances, intersection angles, bicycle lane, delivery type, and comfort index significantly affect the severity (8% to 37%).
-
表 1 变量分类与频数
Table 1. Classification and Frequency of Variable
变量 类别 无受伤 轻伤 重伤 合计 变量 类别 无受伤 轻伤 重伤 合计 骑手特征 性别 女性 82 46 21 149 道路条件 道路类型 支路 363 260 162 785 男性 572 461 291 1324 次干路 195 146 87 428 年龄 ≤30 268 291 172 731 主干路 96 101 63 260 > 30~40 254 153 118 525 入口数 2 195 136 97 428 > 40 132 63 22 217 3 130 101 71 302 外卖类型 日常用品 162 111 65 338 4 329 270 144 743 餐饮 492 396 247 1135 交叉口角度 斜交 305 217 149 671 交通行为 头盔 否 134 92 49 275 正交 349 290 163 802 是 520 415 263 1198 信控 无信号 295 209 142 646 超速 否 331 244 166 交通行为 有信号 359 298 170 827 是 323 263 146 732 路面 干燥 410 309 184 903 闯红灯 否 361 424 213 998 潮湿 244 198 128 570 是 293 83 99 475 自行车专用道 是 197 156 93 446 逆行 否 364 409 247 1020 否 457 351 219 1027 是 290 98 65 453 交通环境 天气 晴朗 236 166 94 496 变道 否 464 336 240 1040 多云 200 143 88 431 是 190 171 72 433 雨/雪 218 198 130 546 时间特征 季节 春季 176 105 74 355 光线 白天 345 256 168 769 夏季 192 148 92 432 黄昏/黎明 132 88 69 289 秋季 159 97 63 319 黑夜 177 163 75 415 冬季 127 157 83 367 风力强度 4级以下 511 381 253 1145 节日 周末 238 176 111 525 其他 143 126 59 328 工作日 416 331 201 948 舒适度 不舒适 295 217 144 656 时间 深夜 43 27 22 92 舒适 359 290 168 817 平峰 352 265 162 779 事故特征 碰撞对象 固定物 103 67 30 200 高峰 259 215 128 602 行人 136 95 41 272 空间特征 区位 城区 445 321 209 975 自行车 128 88 46 262 机动车 287 257 195 739 郊区 209 186 103 498 行程 取货 211 151 109 471 位置 路段 201 141 102 444 配送 443 356 203 1002 交叉口 453 366 210 1029 合计 654 507 312 1473 表 2 模型参数估计与边际效应汇总表
Table 2. Summary of Model Parameter Estimation and Marginal Effects
变量名称 域1 域2 边际效应/% 系数 S.E. 系数 S.E. 无受伤 轻伤 重伤 性别(基于:女性) 男性* -0.221 0.036 0.147 0.032 7.58 2.25 -9.83 年龄(基于:≤30) > 30~40** -0.172 0.016 -0.192 0.018 33.68 -19.85 -13.83 > 40*** -0.153 0.019 0.025 0.017 35.81 -23.27 -12.53 外卖(基于:日常用品)PL 餐饮* 0.094 0.044 0.094 0.044 -9.12 4.19 4.93 头盔(基于:否) 是*** -0.617 0.018 -0.662 0.022 19.63 -6.76 -12.87 超速(基于:否) 是** 0.725 0.009 0.864 0.022 -55.52 12.82 42.71 闯红灯(基于:否) 是*** 1.156 0.013 1.347 0.018 -56.81 6.30 50.52 逆行(基于:否) 是*** 0.871 0.037 0.882 0.032 -51.13 5.67 45.46 变道(基于:否) 是** 0.624 0.021 0.711 0.023 -26.37 9.87 16.50 季节(基于:春季)PL 夏季*** 0.178 0.045 0.178 0.045 -9.77 2.92 6.86 秋季 冬季** 0.149 0.044 0.149 0.044 -7.21 -1.75 8.96 节日(基于:周末) 工作日 时间(基于:深夜) 平峰** 0.613 0.037 -0.523 0.032 20.06 5.46 -25.52 高峰*** 0.684 0.039 -0.416 0.035 23.67 2.62 -26.29 区位(基于:城区) 郊区* -0.068 0.052 0.135 0.046 -5.61 -3.43 9.04 位置(基于:路段)PL 交叉口** 0.357 0.029 0.357 0.029 13.23 -23.51 10.27 道路类型(基于:支路) 次干路** -0.266 0.032 0.294 0.036 -15.22 5.15 10.06 主干路*** -0.314 0.031 0.367 0.028 -17.15 4.54 12.61 入口数(基于:2)PL 3* 0.213 0.028 0.213 0.028 12.66 -2.69 -9.96 4** 0.311 0.033 0.311 0.033 16.45 -2.87 -13.59 交叉口角度(基于:斜交)PL 正交* -0.316 0.017 -0.316 0.017 17.61 -3.07 -14.54 信控(基于:无信号) 有信号*** -1.031 0.021 -1.035 0.022 -12.74 -6.10 18.84 路面(基于:干燥)PL 潮湿** 0.215 0.016 0.215 0.016 -18.62 -3.42 22.04 自行车道(基于:是) 否*** 0.371 0.018 0.405 0.019 -31.08 5.87 25.21 天气(基于:晴朗)PL 多云 雨/雪*** 0.157 0.008 0.157 0.004 -10.60 4.20 6.40 光线(基于:白天) 黄昏/黎明* 0.094 0.033 0.098 0.031 -9.63 -1.71 11.34 黑夜*** 0.116 0.026 0.122 0.028 -11.22 2.68 8.54 风力(基于: < 4级) 其他 舒适度(基于:不舒适)PL 舒适** 0.049 0.016 0.049 0.016 -4.45 8.78 -4.33 行人** -0.077 0.019 -0.092 0.022 20.01 5.92 -25.93 碰撞对象(基于:固定物) 自行车** 0.112 0.022 0.131 0.019 -22.99 8.46 14.53 机动车*** 0.497 0.018 0.556 0.027 -38.79 11.02 27.77 行程(基于:取货) 配送 注:“域1”为“无受伤与轻伤”对比域;“域2”为“轻伤与重伤”对比域;S.E.为标准误差;“—”为不具有统计显著性,“PL”为变量遵循平行线假设;“*”“**”“***”分别为显著水平为P < 0.10,P < 0.05,P < 0.01。 -
[1] 李佳星. 基于扎根理论的外卖骑手交通事故影响因素研究[D]. 成都: 西南交通大学, 2021.LI J X. Research on the factors affecting the traffic accidents of takeaway riders based on the grounded theory[D]. Chen-du: Southwest Jiaotong University, 2021. (in Chinese) [2] 王超, 周璐好, 任倩文. 基于感知失衡的外卖配送交通安全研究[J]. 中国安全生产科学技术, 2021, 17(8): 162-166.WANG C, ZHOU L H, REN Q W. Research on traffic safety of takeaway delivery based on perception imbalance[J]. Jour-nal of Safety Science and Technology, 2021, 17(8): 162-166. (in Chinese) [3] 殷豪, 林淼, 王鹏, 等. 基于潜类别Logit模型的两轮车骑行者头部损伤影响因素研究[J]. 交通信息与安全, 2023, 41 (1): 43-52. doi: 10.3963/j.jssn.1674-4861.2023.01.005YIN H, LIN M, WANG P, et al. An analysis of the impact fac-tors of head injuries of two-wheeler riders using a latent class logit model[J] Journal of Transport Information and Safety, 2023, 41(1): 43-52. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.01.005 [4] 朱新宇, 褚昭明, 朱建安, 等. 中国电动自行车交通事故析因及对策建议[J]. 城市交通, 2021, 19(6): 62-68.ZHU X Y, CHU Z M, ZHU J A, et al. Analysis and counter-measures of electric bicycle traffic accident in China[J]. Ur-ban Transport of China, 2021, 19(6): 62-68. (in Chinese) [5] 杜萌萌. 外卖骑手致交通事故的法律责任问题研究[D]. 扬州: 扬州大学, 2022.DU M M. Study on legal liability of takeout riders in traffic accidents[D]. Yangzhou: Yangzhou University, 2022. (in Chi-nese [6] 张新芬. 信号交叉口外卖电动车穿越危险行为影响因素研究[D]. 西安: 长安大学, 2020.ZHANG X F. Study on the influencing factors of crossing dangerous behaviors of takeaway electric bicycles at signal-ized intersections[D]. Xi'an: Chang'an University, 2020. (in Chinese) [7] 魏宇浩, 秦华. 外卖电动自行车在交叉口的风险行为研究[J]. 北京建筑大学学报, 2021, 37(1): 25-30.WEI Y H, QIN H. Study on risk behavior of take-away elec-tric bicycles at intersections[J]. Journal of Beijing University of Civil Engineering and Architecture, 2021, 37(1): 25-30. (in Chinese) [8] 佀国磊. 外卖骑手交通事故风险分析[J]. 兰州工业学院学报, 2020, 27(5): 87-90.SI G L. Traffic accident risk analysis of takeaway delivery rid-ers[J]. Journal of Lanzhou Institute of Technology, 2020, 27(5): 87-90. (in Chinese) [9] CHAI H, ZHANG Z P, XUE J, et al. A quantitative traffic per-formance comparison study of bicycles and E-bikes at the non-signalized intersections: evidence from survey data[J]. Accident Analysis & Prevention, 2022, 178: 106853. [10] XU C, YANG Y, JIN S, et al. Potential risk and its influenc-ing factors for separated bicycle paths[J]. Accident Analysis & Prevention, 2016, 87: 59-67. [11] YUAN Q, YANG H Q, HUANG J, et al. What factors impact injury severity of vehicle to electric bike crashes in Chi-na?[J]. Advances in Mechanical Engineering, 2017, 9(8): 1687814017700546. [12] DU W, YANG J, POWIS B, et al. Epidemiological profile of hospitalised injuries among electric bicycle riders admitted to a rural hospital in Suzhou: a cross-sectional study[J]. Inju-ry Prevention, 2014, 20(2): 128-133. [13] XU C, GUO H, XU L, et al. Speeding behavior and speed limits for heterogeneous bicycle flow[J]. Traffic Injury Pre-vention, 2019, 20(7): 759-763. [14] WANG X, CHEN J, QUDDUS M, et al. Influence of famil-iarity with traffic regulations on delivery riders'e-bike crash-es and helmet use: two mediator ordered logit models[J]. Ac-cident Analysis & Prevention, 2021, 159: 106277. [15] WANG T, CHEN J, WANG C, et al. Understand e-bicyclist safety in china: Crash severity modeling using a generalized ordered logit model[J]. Advances in Mechanical Engineer-ing, 2018, 10(6): 1687814018781625. [16] 马景峰, 任刚, 李豪杰, 等. 电动自行车与机动车事故严重性影响因素分析[J]. 交通运输系统工程与信息, 2022, 22 (2): 337-348.MA J F, REN G, LI H J, et al. Analysis on contributing fac-tors influencing severity of e-bicycle and motor vehicle crash-es[J]. Journal of Transportation Systems Engineering and In-formation Technology, 2022, 22(2): 337-348. (in Chinese) [17] 赵跃峰, 张生瑞, 马壮林. 基于部分优势比的公路隧道交通事故严重程度分析模型[J]. 中国公路学报, 2018, 31(9): 159-166.ZHAO Y F, ZHANG S R, MA Z L. Analysis of traffic acci-dent severity on highway tunnels using the partial proportion odds model[J]. China Journal of Highway and Transport, 2018, 31(9): 159-166. (in Chinese) [18] WILLIAMS R. Generalized ordered logit/partial proportion-al odds models for ordinal dependent variables[J]. The Stata Journal, 2006, 6(1), 58-82. [19] 吕拉昌, 陈东霞. 人居环境气候舒适度对城市创新的影响分析[J]. 地域研究与开发, 2021, 40(2): 45-49.LYU L C, CHEN D X. Impact of climate comfort of urban human settlements on urban innovation[J]. Areal Research and Development, 2021, 40(2): 45-49. (in Chinese) [20] 王政阳, 丛浩哲, 谢晓颖, 等. 面向快递外卖人员的交通安全精准化宣传教育策略与方法研究[J]. 汽车与安全, 2022 (8): 88-94.WANG Z Y, CONG H Z, XIE X Y, et al. Research on the precise traffic safety publicity and education strategies and methods for couriers and take-away deliverymen[J]. Auto & Safety, 2022(8): 88-94. (in Chinese) [21] 牛莉霞, 韩羲秀, 赵蕊. "赶工游戏"对外卖骑手不安全驾驶的影响: 基于压力认知评价的视角[J]. 中国安全科学学报, 2024, 34(4): 17-25.NIU L X, HAN X X, ZHAO R. Effect of "rush game" on un-safe driving of take-away riders: a perspective based on eval-uation of stress perception[J]. China Safety Science Journal, 2024, 34(4): 17-25. (in Chinese) [22] 蔡凌霄, 周备, 张生瑞, 等. 考虑均值及方差异质性的外卖骑手闯红灯行为影响因素分析[J]. 交通信息与安全, 2024, 42(1): 59-66. doi: 10.3963/j.jssn.1674-4861.2024.01.007CAI L X, ZHOU B, ZHANG S R, et al. Factors affecting red-light running behaviors of takeaway delivery riders con-sidering heterogeneity in the means and variances[J]. Jour-nal of Transport Information and Safety, 2024, 42(1): 59-66. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2024.01.007