USA Banner

Official US Government Icon

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure Site Icon

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

U.S. Department of Transportation U.S. Department of Transportation Icon United States Department of Transportation United States Department of Transportation
FHWA Highway Safety Programs


Step 7 in the scalable risk assessment process consists of compiling other data besides exposure that is required based upon the risk definition selected in Step 3. The three possible risk definitions are:

  1. Observed crash rate
  2. Expected crashes
  3. Additional risk indicators

Detailed instructions for compiling other required data for these three risk definitions is beyond the scope of this guide. There is extensive guidance and examples in several other reports, manuals, and guides. Therefore, the following sections provide summary information and pointers to these other guidance documents.

Observed Crash Rate

To calculate observed crash rate, reported pedestrian and bicyclist crash data are compiled from existing state and local crash databases. The exact procedures for obtaining and compiling crash data vary from state to state (as well as the crash data attributes). Therefore, this section provides an overview and points to other published resources and guide. In particular, these FHWA documents are relevant for compiling crash data:

  • Highway Safety Improvement Program (HSIP) Manual, FHWA-SA-09-029, January 2010.
  • Guidebook on Identification of High Pedestrian Crash Locations, publication pending, June 2018.

Each agency that provides crash data will typically provide documentation and data dictionaries that describe crash database attributes. Typically, a crash database contains three major components:

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

If observed crashes are being used to quantify risk, the possibility of unreported crashes should be considered as a potential bias. In some cases, safety analysts will supplement official crash databases with other sources of data, such as that from emergency medical services, hospital outcomes, and public health databases. Considering these other sources may help to provide a more comprehensive database of pedestrian and bicyclist crashes.

Expected Crashes

Expected pedestrian and bicyclist crashes can be estimated by:

  • Using HSM procedures to estimate point (i.e., intersection) or segment level predicted crashes, then using Empirical Bayes procedures to blend observed crashes and predicted crashes to estimate expected crashes at a specific location.
  • Developing a crash prediction model at aggregate or disaggregate level using observed crashes and other causal factors, then using Empirical Bayes procedures to estimate expected crashes. In some situations, a safety performance function is not available within the HSM and the blending can use locally developed safety performance functions with the observed crashes to estimate expected crashes.

In either case, exposure is considered an important factor in estimating expected crashes. Therefore, the exposure values developed in Step 6 will be used in this step to estimate expected crashes.

Safety analysts estimate expected crashes to overcome several issues associated with observed crashes. Observed crashes (especially pedestrians and bicyclists) can be a rare occurrence, and the actual observed number of crashes at specific locations may not accurately represent the risk to pedestrians and bicyclists.

The HSM has developed safety performance functions that are used to calculate predicted crashes. Then, Empirical Bayes procedures are used to estimate expected crashes (which is a weighted average of observed crashes and predicted crashes). However, at the time of this writing, the HSM procedures for pedestrian and bicyclist crashes are still being refined and are not comprehensive (e.g., they do not address crashes on rural roads). NCHRP Project 17-84 was initiated in early 2017 to improve guidance for pedestrian and bicyclist crash prediction in future editions of the HSM.

Several efforts have developed crash prediction models aside from those in the HSM. The development of crash prediction models is outside the scope of this Guide, but the following list includes examples of crash model development for interested readers:

  • Turner, S., Wood, G., Hughes, T., Singh, R. Safety Performance Functions for Bicycle Crashes in New Zealand and Australia. Transportation Research Record 2236, pp. 66–73, 2011.
  • Pulugurthaa, S., and Sambhara, V. Pedestrian Crash Estimation Models for Signalized Intersections. Accident Analysis and Prevention, Vol. 43, pp. 439–446, 2011.
  • Nordback, K., Marshall, W., and Janson, B. Bicyclist safety performance functions for a U.S. city. Accident Analysis and Prevention, Vol. 65, pp. 114–122, 2014.
  • Alluri, P., Haleem, K., Gan, A., Lavasani, M., and Saha, D. Comprehensive Study to Reduce Pedestrian Crashes in Florida. Florida Department of Transportation, Grant: BDK80 977-32, 2015.
  • Amoh-Gyimah, R., Saberi, M., Sarvi, M. Macroscopic modeling of pedestrian and bicycle crashes: A cross-comparison of estimation methods. Accident Analysis & Prevention, Vol. 93, pp. 147-159, 2016.
  • Thomas, L., Lan, B., Sanders, R., Frackelton, A., Gardner, S., and Hintze, M. Changing the Future? Development and Application of Pedestrian Safety Performance Functions to Prioritize Locations in Seattle, WA. Transportation Research Board 96th Annual Meeting Compendium Papers, Washington D.C., 2017.

Additional Risk Indicators

In this definition of risk, analysts develop and compile additional risk indicators that have been defined in Step 3. The actual risk indicators may vary depending upon the location and facilities being analyzed, and are identified as part of a systemic safety evaluation process (or similar process). FHWA provides several resources for systemic safety at In particular, three documents are relevant:

  • Systemic Safety Project Selection Tool, FHWA-SA-13-019, July 2013
  • Systemic Safety Project Selection Tool Supplemental Case Studies, FHWA-SA-17-002, December 2016
  • Thomas et al. Systemic Pedestrian Safety Analysis, NCHRP Project 17-73, anticipated 2018.

The case studies in Report FHWA-SA-17-002 include an example of systemic analysis for pedestrian and bicyclist crashes, and this example was included earlier in this guide.

Attributes from roadway inventory and traffic count databases are often the starting point for identifying risk factors. For example, FHWA recommends the following list of potential risk factors for consideration in systemic safety analyses.

Roadway and Intersection Features

  • Number of lanes
  • Lane width
  • Shoulder surface width/type
  • Median width/type
  • Horizontal curvature, delineation, or advance warning
  • Horizontal curve and tangent speed differential
  • Roadside or edge hazard rating
  • Driveway density
  • Presence of shoulder or centerline rumble strips
  • Presence of lighting
  • Presence of on-street parking
  • Intersection skew angle
  • Intersection traffic control device
  • Number of signal heads vs. number of lanes
  • Presence of backplates
  • Presence of advanced warning signs
  • Intersection located in/near horizontal curve
  • Presence of left-turn or right-turn lanes
  • Left-turn phasing
  • Allowance of right-turn-on-red
  • Overhead versus pedestal mounted signal heads
  • Pedestrian crosswalk presence, crossing distance, signal head type

Traffic Volume

  • Average daily traffic volumes
  • Average daily entering vehicles

Other Features

  • Posted speed limit or operating speed
  • Presence of nearby railroad crossing
  • Presence of automated enforcement
  • Adjacent land use type, such as schools, commercial, or alcohol-sales establishments
  • Location and presence of bus stops