Scoring Intersection Safety
New indices can help highway engineers create safer environments for pedestrians and bicyclists at intersections.
According to the National Highway Traffic Safety Administration's National Center for Statistics and Analysis, in 2006, 4,784 pedestrians and 773 bicyclists were killed in crashes involving motor vehicles. Collisions at intersections account for 1,008 of those pedestrian fatalities and 248 bicyclist fatalities, indicating that pedestrian and bicyclist safety remain major issues that require continued focus.
Many States and municipalities have safety programs to identify and address locations with high numbers of crashes involving pedestrians and bicyclists. Some programs treat hazards only after they become apparent; too often, after someone is hurt. To reduce the potential for fatalities and injuries, proactive methods for prioritizing intersections for upgrades in signage, signal timing, and other controls are highly preferable.
Road safety audits (RSAs) have proven successful in many States, as described in the article "RSAs for Safety" in the November/December 2006 issue of Public Roads. In fact, the Federal Highway Administration's (FHWA) Office of Safety recently released the Pedestrian Road Safety Audit Guidelines and Prompt Lists (http://drusilla.hsrc.unc.edu/cms/downloads/PedRSA.reduced.pdf), which provides transportation agencies and teams conducting RSAs with a better understanding of the needs of pedestrians of all abilities.
In addition, FHWA's Pedestrian Safety Guide and Countermeasure Selection System (also known as PEDSAFE and BIKESAFE, available at www.walkinginfo.org/pedsafe and www.bicyclinginfo.org/bikesafe) offer online tools for engineering, education, and enforcement treatments that can improve pedestrian and bicyclist safety and mobility at user-defined locations.
Two new tools that could help States gauge traffic safety, specifically for intersections, are FHWA's Pedestrian Intersection Safety Index (Ped ISI) and Bicycle Intersection Safety Index (Bike ISI). The ISIs identify and prioritize potentially dangerous intersections — that is, before an injury or death has necessarily occurred there — enabling safety engineers and other practitioners to study the intersections for possible amendment.
"The Ped and Bike ISIs use observable, easy-to-gather data, making the ISIs feasible to be applied on a wide scale," says Michael Trentacoste, director of FHWA's Office of Safety Research and Development. "We anticipate that States, localities, consulting engineers, and other interested parties will take advantage of the ISIs to help them prioritize intersections for further study."
Adds FHWA Senior Highway Safety Engineer Rudy Umbs, "The indices are key tools that States can use as they develop and implement their Strategic Highway Safety Plans and Pedestrian Safety Action Plans."
A Persistent Problem
Urban areas tend to have the highest rates for both pedestrian and bicyclist crashes with motorized vehicles. Around one-half of pedestrian injuries related to crashes involving motor vehicles occur at intersections, such as roadways, driveways, and alleys. A variety of factors play a role, including pedestrian age, width of the crossing, street corners with large turning radii that permit high motor vehicle speeds, and misunderstanding of pedestrian signals.
For bicyclists, more than one-half of injuries related to crashes involving motor vehicles occur at intersections. Related factors include the age of the bicyclist, vehicle speeds and traffic volumes, presence of auxiliary right-turn lanes, and other designs that lead to weaving between bicycles and motor vehicles.
"Given the desire to avoid crashes involving pedestrians and bicyclists at all intersections, FHWA felt that an efficient and effective approach was needed to locate not only high crash locations but all intersections that had the potential to be problem crossings," says Trentacoste. "States and localities have adopted road safety audits and other means of deciding which intersections need safety improvements, and what we came up with can be applied without having to wait for pedestrian or bicycle crashes to occur before making a cost-effective change."
The ISIs enable users to identify any intersection crossings (for pedestrians) and approach legs (for bicyclists) that should be priorities for safety improvements. Using observable characteristics of crossings and approach legs, the ISIs produce safety index scores, with the high scores indicating greater priority for indepth safety assessments.
Each leg of an intersection can have different characteristics affecting pedestrian or bicyclist safety; therefore, the ISIs are intended to provide evaluations of individual crossings or approach legs rather than evaluating an intersection as a whole. A safety engineer can use the ISIs to develop a prioritization scheme for a group of pedestrian crossings or bicyclist approaches.
Some practitioners already are having success. "We recently used the Intersection Safety Indices to assess approximately 200 pedestrian crossings in a growing community in Virginia," says Dan Nabors, a senior transportation engineer with engineering contractor Vanasse Hangen Brustlin, Inc., which is supporting a project in Fairfax County. "This application helped us target crossings for more detailed assessments and establish priorities for improvements of crossings."
The Ped and Bike ISIs are available as parts of the Pedestrian and Bicyclist Intersection Safety Indices Final Report (FHWA-HRT-06-125, www.fhwa.dot.gov/publications/research/safety/pedbike/06125/) and Pedestrian and Bicyclist Intersection Safety Indices: User Guide (FHWA-HRT-06-130, www.fhwa.dot.gov/publications/research/safety/pedbike/06130/index.cfm). Both documents include links to downloadable Microsoft® Excel® spreadsheet files for ease of using the ISIs. After keying in responses to the various characteristics, users receive an ISI value that rates the intersection by potential danger.
Developing the ISIs
The FHWA researchers who developed the ISIs studied 68 pedestrian crossings at signalized and unsignalized intersections in urban and suburban areas of Miami, FL; Philadelphia, PA; and San José, CA. They also studied 67 bicycle approaches at signalized and unsignalized intersections in urban and suburban Eugene, OR; Gainesville, FL; Philadelphia, PA; and Portland, OR.
The intersections had the following characteristics:
- Three-leg or four-leg intersections
- Signalized two-way or four-way stop
- Traffic volumes of 600 to 50,000 vehicles per day
- One-way or two-way roads
- One to four through lanes
- Speed limits of 24 to 72 kilometers per hour, km/h (15 to 45 miles per hour, mi/h)
Traditionally, the most commonly used measure of safety for a site is crash history. When dealing with intersections, however, crash data normally are gathered for an intersection as a whole. But the ISI research required more specific data because the unit of study was a single crosswalk (for pedestrians) or single approach (for bicyclists).
Initially, State and city departments of transportation provided lists of crashes involving pedestrians or bicycles for each study site in their jurisdictions. In most cases the crash information database did not have sufficient location information to pinpoint the exact part of the intersection where the crash occurred. The FHWA researchers then obtained copies of police-recorded crash reports and examined the sketches and narratives. The crashes involving pedestrians or bicyclists were noticeably few. In general, at any given location, pedestrian and bicyclist crashes are rare events, made even rarer in this study because only a portion of the crashes at an intersection were considered. (That is, a site was defined as one intersection leg. Thus, if the crashes were spread evenly around the intersection, only about one-quarter of the total intersection crashes would be considered.)
The FHWA researchers focused on two measures of safety: safety ratings by expert evaluators and observed interactions between pedestrians and motorists or bicyclists and motorists. They selected these two measures because limited crash data were available. For the first measure, the researchers enlisted people knowledgeable in pedestrian and bicyclist matters — such as engineers, planners, pedestrian and bicycle coordinators, and advocates for the visually impaired and people with other disabilities — and obtained their opinions on the sites. To provide enough information for the evaluators to develop safety index ratings, the researchers created a Web-based survey instrument consisting of an illustration and video clip for each site. The evaluators were asked to view each illustration and video as if they were pedestrians crossing at the crosswalk or bicyclists approaching the intersection.
The evaluators rated the sites on a scale of one to six according to their sense of safety and comfort. If conditions made them feel very comfortable and highly likely to walk or ride at the site, they were instructed to give a rating of one. If conditions made them feel very uncomfortable and highly unlikely to walk or ride at the site, they were instructed to give a rating of six. They also had the option to cite "not enough information" if an illustration or video did not give them a good feel for an area. Each evaluator in the pedestrian safety survey gave a single rating per crosswalk. Each evaluator in the bicyclist survey gave a single rating for each movement that a bicyclist could make at the intersection (straight through, left turn, and right turn).
For the second measure of safety, the researchers videotaped each site over long periods to record interactions between pedestrians and motorists or bicyclists and motorists. The study observed a total of 4,128 pedestrian events over 90 hours and 3,831 bicyclist events over 129 hours.
The researchers later watched the videos and coded conflicts and avoidance maneuvers as they occurred. Recorded behavior included changes in speed or direction by a pedestrian, bicyclist, or motorist in response to the presence of another party. In total, the researchers recorded 1,898 bicyclist-motorist interactions and 1,095 pedestrian-motorist interactions.
Crunching the Numbers
After collecting the safety measures, the researchers developed models. Those based on the evaluator ratings used multiple linear regressions, as the ratings generally followed a normal distribution. The models based on the recorded behavior used a generalized linear model, as the behavioral data generally followed a statistical probability model called the Poisson distribution.
The ratings-based models served as the core of development of the Ped ISI and Bike ISI, although the behavior-based models also contributed to both. The fact that the ratings-based models predict safety ratings for sites on a scale of one to six conveniently led to development of a safety index.
The Ped ISI model consists of one equation that determines the safety index score for a single pedestrian crossing. The model indicates an increasing priority for intersection legs that have no signal or stop-sign control, more through travel lanes, higher vehicle speeds, higher traffic volumes where signals are present, and predominantly commercial surrounding land uses.
The Bike ISI consists of three equations, each one determining the safety index score for a single bicycle movement (straight through, right turn, or left turn) at a single intersection approach leg. The ISI model for through bicycle movements indicates an increasing priority for intersection approaches with higher traffic volumes, speed limits greater than 56 km/h (35 mi/h), presence of turning vehicle traffic across the path of through cyclists, combination of right-turn lanes and bicycle lanes, higher cross-street traffic when there is no bicycle lane, signalization when there is no bicycle lane, and onstreet parking.
The model for right-turn bicycle movements indicates an increasing priority for intersection approaches with higher traffic volumes, more lanes for the bicyclist to cross to make the turn, more lanes on the cross street, and onstreet parking.
The model for left-turn bicycle movements indicates an increasing priority for intersection approaches with higher traffic volumes, bicycle lanes, signalization, speed limits greater than 56 km/h (35 mi/h), more lanes for the bicyclist to cross to make the turn, and onstreet parking.
Using the ISIs
The steps for using the ISIs are as follows:
Select sites to evaluate. Identify pedestrian crossings or intersection approaches to evaluate. Although the ISIs are intended to rank problem sites on a townwide or citywide basis, it might be more immediately helpful to start instead with sites that have higher volumes of pedestrian or bicyclist activity, areas that have been a concern to the community, or intersections that are still in the planning stage.
Gather data. Gather data on geometric and operational characteristics of the selected sites, either through electronic databases or brief field visits. If the sites are in the planning stages, determine what characteristics the sites are expected to have.
Calculate index values. Use the ISIs to produce safety index values for each site, which will receive a value between 1 (safest) and 6 (least safe). The calculation can be performed by using the formulas, quick reference tables, or Excel spreadsheet calculator that accompany FHWA's final report and user guide.
Prioritize sites. Sort sites according to safety index values. The ISI value for a site will indicate its relative priority for safety assessment. A site with an ISI value of 1.5 would probably have low speeds, low traffic, few travel lanes, and be located in a residential area. A site with a value of 3.0 would have moderate speeds and traffic volumes and an exclusive right- or left-turn lane. A site with a value of 4.5 would have high traffic volumes and speed, multiple lanes, and high levels of turning traffic with many turn lanes. Although these are general representations of probable ISI values for certain types of sites, the user must remember that this is a prioritization of sites. No matter what the particular ISI value, sites with the highest values generally have the highest priority for further evaluation of pedestrian and/or bicyclist safety.
However, the existence of a high Ped ISI or Bike ISI value alone might not always provide a complete depiction of a crosswalk's or intersection's hazard level. Many characteristics and behaviors at an intersection can result in a pedestrian or bike crash, and no model can include all these factors. The practitioners' knowledge of the locations also should be used in prioritizing sites.
The following section illustrates how to apply the ISIs, using a pedestrian crossing at an intersection as an example. The crossing location has the following characteristics:
- Signalized intersection
- No stop control
- Four through lanes on the main road (two in each direction)
- The 85th percentile speed, which is the speed driven by 85 percent of motorists, on the main road is 68 km/h (42 mi/h)
- Main road average daily traffic (ADT) is 22,000 vehicles
- Surrounding area is residential
In this example, the user would access the Ped ISI Excel spreadsheet and type "1" into the SIGNAL box (as opposed to "0" for an unsignalized intersection), type "0" into the STOP box (as opposed to "1" for the presence of stop control), type "4" into the THRULNS box, type "42" into the SPEED box, type "22,000" into the MAINADT box, and type "0" into the COMM box ("0" for residential rather than a commercial, "1," area). Then the ISI automatically would calculate the safety index value: 2.7, moderate on the 1-6 scale.
Another way to determine the Ped ISI value is to go directly to the quick reference tables of values. These tables are provided for various combinations of traffic control and area type. For the example above, the user would find the table corresponding to a signalized crossing in a noncommercial area. Looking at that table, with four through lanes, 85th percentile speed of 68 km/h (42 mi/h) (use the column of the nearest value, 40 mi/h), and 22,000 ADT (use the row with 20,000 ADT), the Ped ISI value is 2.7.
A First Step
The ISIs enable practitioners to evaluate each approach at all intersections, or a select group of intersections, in their jurisdictions. The prioritization approach used in the ISIs helps users target the most dangerous sites and stay within budgetary confines. Once safety index values are assigned to each site, a practitioner can conduct more detailed reviews of the sites with the highest values, using other tools and methods such as safety audits to determine whether any geometric or traffic control treatments are needed to improve safety.
"Overall, I applaud what FHWA has accomplished with its ISIs, and recognize the difficulty of reducing the myriad factors affecting bicycle safety into an equation with coefficients that can be reliably estimated from a limited sample," says Peter Furth, professor of civil and environmental engineering at Northeastern University and member of the Transportation Board in Brookline, MA.
Ann H. Do is a research highway engineer at FHWA's Turner-Fairbank Highway Research Center in McLean, VA, where she has managed the Pedestrian and Bicycle Safety Research Program since 2001. She joined FHWA in 1990 as a highway design engineer with the Eastern Federal Lands Highway Division. Do specializes in research related to safety effectiveness evaluations, pedestrians, bicyclists, human factors engineering, and geometric design. She has a B.S. in civil engineering from Virginia Polytechnic Institute and State University (Virginia Tech).
Daniel Carter received B.S. and M.S. degrees in civil engineering (transportation) from North Carolina State University. He is an engineering research associate at the Highway Safety Research Center (HSRC) at the University of North Carolina, where he handles data collection and analysis for safety research projects.
Charles Zegeer is director of the Pedestrian and Bicycle Information Center at HSRC. He has been principal investigator on numerous safety studies, including pedestrian and bicyclist safety. He holds a bachelor's degree from Virginia Tech and a master's degree from the University of Kentucky, both in civil engineering.
William Hunter is a senior research scientist at HSRC. He has been a principal investigator on many Federal studies, most recently pertaining to bicyclist and pedestrian safety. He received B.S. and M.S.C.E. degrees in civil engineering (transportation) from North Carolina State University.
Hardcopies of the ISIs are available through the National Technical Information Service at www.ntis.gov. For more information, see the FHWA TechBrief Pedestrian and Bicyclist Intersection Safety Indices (FHWA-HRT-06-129) at www.fhwa.dot.gov/publications/research/safety/pedbike/06129/, or contact Ann Do at 202-493-3319 or email@example.com; Daniel Carter at 919-962-8720 or firstname.lastname@example.org; or Charles Zegeer at 919-962-7801 or email@example.com.