Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
CHAPTER 1 Introduction The last decade has seen tremendous growth in resources available to state and local transportation agencies for conducting data-driven safety analysis (DDSA). As one major example, the American Association of State Highway and Transportation Officials (AASHTO) published the First Edition of the Highway Safety Manual (HSM) in 2010. The HSM includes methods, procedures, and tools to quantify safety performance. The desired impact of the HSM is more explicit consideration of the expected, substantive safety performance outcomes (typically measured as average crash frequency by different crash types and severity levels) of decisions made during highway and street planning, programming, design, construction, operations, and maintenance. Various stakeholder groups, including AASHTO, the Federal Highway Administration (FHWA), state departments of transportation (DOTs) leading HSM implementation, and Transportation Research Board (TRB) standing committees, have cooperated in efforts to advance use of the HSM to make data-driven decisions, further develop effective road safety management approaches, and ultimately reduce traffic fatalities and serious injuries. Part D of the HSM First Edition consists of crash modification factors (CMFs), multiplicative factors used to estimate the expected change in the long-term average crash frequency after implementing a given treatment at a site. A treatment can be any change affecting a site. Some treatments are implemented to improve safety performance. Some treatments are implemented to address other performance objectives and needs (e.g., improve travel time reliability), but it is still desirable to know if and how the treatment will affect safety performance (e.g., improve safety, reduce safety, no effect on safety). Developing CMFs typically requires information about crash frequency and severity, traffic volumes, and road characteristics, including where and/or when the treatment of interest was implemented. This information must be available both before and after implementation of a treatment or at sites with and without the treatment. While basing CMFs on expected changes in crash outcomes is ideal, there are some challenges associated with relying only on crash data for quantifying safety performance. First, CMF developers usually obtain crash data from police crash reports, and these are subject to reporting thresholds and error or uncertainty about some crash details. Additionally, multiple years of crash data are typically aggregated for analysis, which generally assumes static conditions over that period. This has made it difficult to develop CMFs for new and innovative designs and strategies for which multiple years of data are not available, or for specific crash and facility type combinations that tend to have relatively smaller sample sizes (e.g., lane departure on low-volume roads). It has also made estimating CMFs challenging for contexts that are dynamic in nature, such as work zones and a wide array of dynamic operational strategies that vary with traffic and weather conditions. Finally, CMFs for treatments seeking to improve pedestrian, bicyclist, and motorcyclist safety have been elusive because of the need for multiple years of crash and exposure data, the latter of which is rarely available for these users. Pedestrian and bicyclist CMFs remain a significant gap in 1
the HSM, with multiple efforts to partially fill this gap occurring under the National Cooperative Highway Research Program (NCHRP). Based on current draft text, Part D of the HSM Second Edition will speak to the potential use of alternative, or surrogate, measures of safety to evaluate treatments and estimate CMFs. Surrogate measures of safety were identified by researchers nearly 40 years ago as quantifiable observations that can be utilized to replace or supplement crash records (Datta et al., 1983). Example surrogate measures include traffic conflicts, lane departures and/or encroachments, traffic control compliance, steering behavior (e.g., yaw rate, lateral acceleration) and stopping behavior (e.g., maximum decelerations). This project takes the approach that surrogate measures could also include macroscopic traffic-level measures (e.g., speed variance, average density), as well as combined quantitative and qualitative assessments and metrics characterizing the road environment, such as design consistency evaluations like those in FHWAâs Interactive Highway Safety Design Model (IHSDM) and intersection evaluations like those in FHWAâs A Safe System-Based Framework and Analytical Methodology for Assessing Intersections (Porter et al., 2021). Surrogate measures of safety are a thriving area of research, but they have not yet routinely made their way into practical procedures for conducting data- driven safety analysis. Surrogate measures can, in theory, be used to estimate a CMF indirectly in lieu of using crash data. The use of surrogate measures to estimate the safety performance effect of one or more treatments is identified as one possible study approach in FHWAâs A Guide to Developing Quality Crash Modification Factors (Gross et al., 2010). Tarko et al. (2009) pointed out that two conditions must be met before a surrogate measure of safety can be used in this way: ï· The surrogate measure should be based on an observable non-crash event that is related to crashes. ï· There should be a practical method for converting the changes or differences in non-crash events into corresponding changes or differences in crash frequency (possibly by crash type or severity). In other words, the key to the application of surrogate measures of safety to estimate CMFs is the availability of reliable models or procedures to relate the surrogate measures, or changes in the surrogate measures to crash frequency (by crash type and severity), or changes in crash frequency. Establishing links between surrogate measures and crashes, particularly for the wide range of surrogates that exists, and across the different site types and contexts of interest, has been a challenging endeavor. Gaps remain. However, there are increasing opportunities for progress with the availability of emerging technologies and new data sources. In the meantime, researchers and practitioners have used and continue to use surrogate measures of safety to evaluate treatments. They use surrogate measures even if established relationships to crashes do not yet exist. For example, evaluations of new traffic control devices (TCDs) under efforts such as the TCD Pooled Fund Consortium (with participation from nearly 30 states) often rely on surrogate measures of safety collected from indoor driving simulators and field studies. Safety-related efforts that leverage the Second Strategic Highway Safety Program (SHRP2) Naturalistic Driving Study (NDS) data are also based largely on surrogate measures. Prior to the completion of robust, crash-based evaluations, early evaluations of what now have become FHWA Proven Safety Countermeasures, such as rumble strips and wider edge lines, were based on surrogate measures such as average lane position and variance in lane position, even though established crash linkages for those surrogates did not exist. In these examples where a crash linkage does not exist, changes in the surrogate measures due to a treatment cannot be translated to expected 2
changes in crash frequency, type, or severity. However, the surrogate measures still provide insights to whether the treatment results in desirable changes in user or traffic-level behavior. Such results provide initial evidence that a treatment may be worth further testing, implementation, and evaluation. Overall Study Objectives The objective of this project was to develop a guide for using alternative, or surrogate, measures of safety for developing CMFs and other quantifiable measures in the absence of crash data. The guide, published as NCHRP Research Report 1069: Estimating Effectiveness of Safety Treatments in the Absence of Crash Data: A Guide, establishes a vision for using surrogate safety measures in practice, and provides information on a range of surrogate measures, data collection methods, and study design principles. The guide supports informed decision-making during project planning, project development, and other road safety management activities. The guide also includes six case studies. This report documents the research activities and findings that resulted in the guide. Scope of Final Report This research report consists of the following chapters: Chapter 2 reviews relevant literature. Chapter 3 describes and assesses different methods for the collection of surrogate measures of safety. Chapter 4 prioritizes types of treatments with no CMFs or low-rated CMFs with respect to evaluation needs using surrogate measures of safety. Chapter 5 presents two detailed case studies that the research team conducted during this project. The case studies demonstrate how surrogate measures of safety can be used to evaluate two countermeasures that fit the high-priority countermeasure categories identified in Chapter 4. Chapter 6 details the development of the guide for using surrogate measures of safety. Chapter 7 summarizes the report and provides future research recommendations. 3
