Guidebook on Identification of High Pedestrian Crash Locations

CHAPTER 3. GATHER DATA

OVERVIEW

Roadway safety has been characterized as nominal or substantive. Nominal safety is based on design standards, while substantive safety is based on roadway safety performance. A roadway may be thought to be nominally safe because it meets minimum design criteria; however, it could have higher-than-expected crash experience. The reverse could also be true, where a roadway is not meeting minimum design criteria and yet has a high level of substantive safety. Substantive safety requires an evidence-based approach to estimate the expected safety of a roadway through data and analysis rather than focusing solely on standards. Making decisions with an evidence-based approach underscores the need for quality data and data systems.

DATA

Safety data can be grouped into three broad categories, as shown in figure 2. The following are two types of data primarily needed for a traditional (reactive) safety study:

• Crash data, including severity, crash type, characteristics of interest (e.g., age of pedestrian or light condition), and contributing factors. The location of the crash, preferably as latitude and longitude coordinates, is also needed.
• Roadway characteristics, such as number of lanes or traffic control devices present.

For a pedestrian safety study, the desired exposure data that can improve the process could include the following:

• Vehicle counts.
• Pedestrian counts.
• Turning and crossing movement counts for specific locations.

Crash Data

Generally, crash data are available from State DOTs, State highway patrols, another public agency, or even universities, varying from State to State. Typically, the data are coded and stored in a relational database structure, with separate tables for the crash, unit, and person information. These tables are linked through identifier fields in each table. Before storing and making the final version of the data available for use, DOTs conduct a quality control check that addresses inconsistencies or omissions, as submitted by the investigating law enforcement agency; assigns final coordinates to crashes; and supplements or enhances the crash report based on a review of the crash narrative. Crash data are stored in a centralized database after data have been collected by law enforcement and first responders and submitted to the State.

©Texas A&M Transportation Institute.

Figure 2. Graphic. Types of safety data.

Data Dictionaries

Data dictionaries for the crash data provide a description of each data variable in the crash database. The database typically consists of the following three levels of data:

• Crash-level datasets contain information about the entire crash, such as crash location, crash date, total fatalities in the crash, and light level.
• Vehicle- or unit-level datasets contain information about each vehicle (or unit) in the crash, such as vehicle type and harmful events. Pedestrians and bicyclists (pedalcyclists) are included as nonmotorized vehicles.
• Person-level datasets contain information about all people in crashes, such as age and belt usage. The dataset includes one record for each person involved in the crash.

Improvements in Data Collection

The data dictionary for the crash database can provide details needed to interpret the crash data.

Electronic crash reporting is increasingly being used in various States as the preferred method of collecting data at crash scenes. This approach allows for the integration of in-car mapping technologies, which avoid the need to postprocess the location of the crash by later adding a Global Positioning System reading, street address, or distance from an intersection. Instead, in-car mapping allows an officer to click on the crash location on a digital map and have all the necessary location data automatically entered into the crash report.

Additional information on crash data collections is included in the Highway Safety Improvement Program (HSIP) Manual.⁽¹²⁾

Roadway Data

The roadway characteristics data may be available from a variety of sources, including electronic databases of roadway features and traffic characteristics, aerial photographs and street views, video logs, and traffic control device inventories. The roadway characteristics could also be collected as part of a field review of the road system at a higher cost. Roadway characteristics might also be collected in a virtual environment with field confirmation, as appropriate.

Many State DOTs and some cities use linear referencing, allowing roadway attributes, such as shoulders or speed limit, to be stored individually and to be defined by the route, along with the start and end points. Crashes can also be referenced in the same system.

Exposure Data

In the safety analysis context, exposure is a measure of the number of potential opportunities for a crash to occur. For motor vehicles, exposure is typically quantified using vehicle miles traveled (VMT), which is the product of an average annual daily traffic (AADT) count and the road segment length for which the count is applicable. For pedestrians, facility-specific exposure has been quantified in many ways, including the following:

• Average pedestrian volume.
• Sum of entering flows (pedestrian and motor vehicle) at intersection.
• Product of pedestrian volume and motor vehicle volume.
• Estimated number of street crossings.
• Estimated total travel distance, in person-miles of travel.
• Estimated total travel time, in person-hours of travel.

Exposure is most often used as a normalization factor (i.e., denominator) to equalize for differences in the quantity of potential crash events in different road environments. Depending on the type of safety analysis, it may or may not be desirable to include exposure. When the number of expected or observed crashes is normalized by exposure, the result is what many consider the individual risk to a pedestrian. This quantity may be useful in some comparisons or analyses. However, in other analyses, the total number of pedestrian crashes or crash frequency is better suited.

Data availability may determine the extent to which exposure can be included in the safety analysis. For example, if exposure data are not readily available, then crash frequency (e.g., number of crashes within a segment for a year) rather than crash rate (e.g., crashes per mile per year) would be the performance measure used. Volume or count data are the primary input data for both motor vehicle and pedestrian exposure, and some exposure measures require supplemental data. For example, segment length is needed for the miles traveled measure, and average travel time or distance is required for person-hours of travel.

Motor vehicle traffic volume data are available for major roads and streets from State DOTs, typically in the form of AADT. Increasingly, State DOTs are also working with city and regional planning staff to routinely collect motor vehicle traffic volumes on lower functional class roads and streets.

Pedestrian count data are less commonly available than motor vehicle count data, but increasingly, many local, regional, and State agencies are routinely collecting pedestrian count data at selected key locations. In some cases, models are being used to estimate pedestrian counts everywhere in a city or region. There are several key resources, including the following, that are intended to promote the widespread collection and estimation of pedestrian count data and, thus, should be consulted if necessary:

• Chapter 4, “Traffic Monitoring for Non-Motorized Traffic” of FHWA’s Traffic Monitoring Guide, available at https://www.fhwa.dot.gov/policyinformation/tmguide/.⁽¹³⁾
• NCHRP Report 797, Guidebook on Pedestrian and Bicycle Volume Data Collection: http://www.trb.org/Main/Blurbs/171973.aspx.⁽¹⁴⁾
• FHWA-HEP-17-012, FHWA Bicycle-Pedestrian Count Technology Pilot Project: Summary Report, available at: https://www.fhwa.dot.gov/environment/bicycle_pedestrian/countpilot/summary_report/.⁽¹⁵⁾
• NCHRP Report 770, Estimating Bicycling and Walking for Planning and Project Development: A Guidebook: http://www.trb.org/Publications/Blurbs/171138.aspx.⁽¹⁶⁾
• FHWA-SA-17-041, Synthesis of Methods for Estimating Pedestrian and Bicyclist Exposure to Risk at Areawide Levels and on Specific Transportation Facilities.⁽⁵⁾

SUPPLEMENTAL DATA SOURCES

Supplemental data that some agencies would like to use in their safety evaluation process include the following:

• Citizens’ observations of locations needing attention, either due to maintenance needs or because of a safety concern.
• Law enforcement observations.
• Trauma center data.
• Land use/development plans.

In some cases, a location for consideration of improvements could be identified through the knowledge that a particular development type is being built or based on observations made by local law enforcement or from citizens’ comments or complaints. School districts may contact the city requesting assistance with a crossing near their school. Cities with traffic-calming programs may also identify locations of concern for pedestrians based on evaluations of drivers’ operating speeds. A sidewalk and curb ramp inventory is another city effort that could identify locations of concerns. Although these techniques are not dependent on crashes, they are examples of identifying locations with the potential for pedestrian crashes and can provide an earlier indication of locations where additional evaluation is needed.

ANALYSIS PERIOD

Number of Years in Analysis

Crashes are random events and naturally fluctuate over time at a given location. The fluctuation over time can make it difficult to determine whether a location has a true safety concern or if it is part of a natural fluctuation. The statistical situation known as regression to the mean (RTM) describes a condition in which crash rates are artificially high during a given period and would be reduced in a later period even without an improvement to the location. Several proactive methods are available to address the RTM concern, including using safety performance functions or a systemic safety analysis.

When using a traditional (reactive) approach, having several years of crash data can also help agencies minimize the likelihood of selecting sites that have a high crash frequency due to being on the higher side of a natural fluctuation. A tradeoff with using a large number of years is that roadway conditions, land use, and/or travel patterns could have changed. In a rapidly developing city, only a few years of data may exist for a site with a specific set of roadway, roadside, or traffic control device characteristics. Having quality geometric and traffic control device data can help to associate crash trends with particular conditions.

The report How to Develop a Pedestrian Safety Action Plan states: “When identifying and prioritizing high crash locations, 3 to 5 years of computerized crash data should be used. For prioritizing corridors or other targeted areas, 1 to 3 years of pedestrian data are acceptable.”⁽¹⁷⁾

Number of Years to Repeat Analysis

The frequency that a pedestrian safety analysis is updated is typically between 1 and 3 yr. For communities with mature roadway and pedestrian facilities and without the implementation of a major safety treatment program, the crash data trends may only have minimal changes from year to year, resulting in a reasonable analysis update of every 3 yr. For communities with changes in growth or that are implementing a major safety program, updating the analysis annually can help to illustrate the effects of growth or the safety program.