U.S. Traffic Safety - A National Crisis: Mitigating Fatality Risk with Objective Decision-Making

与大多数其他学科相比，国家和地方交通安全缺乏标准，规范或广泛接受的程序，可以自信地预测性能。目前，决策者必须依赖于主观评价，而这种评价往往偏重于意见和轶事证据。虽然没有人能说这种做法是多方面交通安全问题的唯一根源，但统计数字表明，目前的做法需要加以研究和改进。在过去的40年里，每年有超过40，000人死于美国的道路。 本研究旨在通过引入主动的协作决策过程，为处理交通安全背景下的不确定性和风险提供客观的分析基础。我们的目标，为地方和国家司法管辖区，是（1）开发一个分析模型的采矿碰撞和交通数据，（2）开发一个公路安全手册（HSM），一个全面的目录的程序和信息进行基于性能的碰撞研究，和（3）带来客观的分析到公路安全改进计划（HSIP）的决策过程。两项关键活动是开发碰撞分析模型和程序，以（a）自信地识别危险道路和（B）准确预测处理现场的性能。统计学习技术将用于将道路分类为有用的类别，并校准、评估和选择可用于规划最安全道路的模型。安全标准和故障概率模型-未能满足标准-将是基于性能的碰撞分析过程的一个组成部分，并应成为有效的风险沟通手段，特别是在该过程需要公众参与的情况下。最终，我们的目标是一个适用于美国所有交通领域的工作HSIP决策过程，根据其优点接受-一个满足主动和协作决策过程的需求，并满足国家的巨大需求。成功过渡到我们的HSIP流程取决于利益相关者的采用。因此，第三项关键活动包括通过举办三个讲习班加强和扩大具有不同技术背景、经验和理念的利益攸关方之间的伙伴关系：（1）利益攸关方对道路安全的看法，（2）风险分析观点，（3）建立战略关系。

In contrast to most other disciplines, national and local traffic safety lacks standards, codes or widely accepted procedures that can confidently predict performance. Currently, decision-makers must rely on subjective evaluation that tends to be heavily weighted toward opinion and anecdotal evidence. While no one can claim that this practice is the sole source of a multifaceted traffic safety problem, statistics suggest that current practice needs to be studied and improved. Over 40,000 people have died on U.S. roadways each year for the past forty years. This research aims to provide an objective, analytic foundation for handling uncertainty and risk in the context of traffic safety by introducing a proactive, collaborative decision-making process. Our goals, for both local and national jurisdictions, are (1) to develop an analytical model for mining crash and traffic data, (2) to develop a Highway Safety Manual (HSM), a comprehensive catalog of procedures and information for conducting performance-based crash studies, and (3) to bring analytic objectivity into the Highway Safety Improvement Program (HSIP) decision-making process. Two key activities are to develop crash analysis models and procedures that can (a) confidently identify hazardous roadways and (b) accurately predict performance at treated sites. Statistical learning techniques will be used to classify roadways into useful categories and to calibrate, assess and select models that can be used for planning roadways with maximum safety. A safety standard and a failure probability model -- the failure to satisfy the standard -- will be an integral part of the performance-based crash analyses process and should be an effective risk communication device, especially when the process requires public participation. Ultimately, our goal is a working HSIP decision-making process applicable to all traffic areas in the United States, accepted on its merits -- one that meets the demand for a proactive and collaborative decision-making process and satisfies a great national need. A successful transition to our HSIP process hinges on stakeholder adoption. Therefore, the third key activity consists of strengthening and expanding partnerships among stakeholders with different technical backgrounds, experience, and philosophies by conducting three workshops: (1) stakeholder perceptions on road safety, (2) risk analysis perspectives, and (3) forming a strategic relationship.