In the spring of 2001, the Federal Highway Administration (FHWA) issued a Request for Applications (RFA) to select one or more local jurisdictions to demonstrate and evaluate the effectiveness of a comprehensive pedestrian safety countermeasures program. As a result, FHWA awarded three cooperative agreements to the following locations: Las Vegas, Nevada; Miami-Dade County, Florida; and San Francisco, California. The three study teams were charged with demonstrating and evaluating the effectiveness of a combined pedestrian safety engineering and intelligent transportation systems (ITS)-based area-wide countermeasures program for reducing pedestrian fatalities, injuries, conflicts, and other surrogate measures of safety.

Each of the field teams conducted two-phase studies. Phase I involved a detailed analysis of pedestrian crashes, the selection of appropriate countermeasures, the development of implementation and evaluation plans, and collection and analysis of baseline data. Phase II involved the actual implementation and assessment of the impacts of the countermeasures identified in Phase I. The project included self-evaluations conducted by each of the field teams, as well as an independent national evaluation and cross-cutting study conducted by an independent contractor.

As a result of the pedestrian safety analyses conducted in Phase I, each team selected a number of pedestrian safety countermeasures for deployment. Throughout the project, some of the selected countermeasures changed due to issues with vendors, procurement, or approval from location jurisdictions to install the countermeasures. Nevertheless, in the end, a wide range of traditional and ITS-based countermeasures were deployed at a large number of sites in the three locations.

EVALUATION OBJECTIVES

The objectives of the evaluations were to assess the safety and mobility impacts of the pedestrian safety countermeasures selected for deployment. The evaluations involved collecting and analyzing quantitative data related to the safety and mobility impacts of the countermeasures.

The field teams collected and compared baseline and post-deployment data at the sites where countermeasures were deployed. A wide range of data was collected, depending on the countermeasures being deployed. Data collected included those associated with safety surrogate measures of effectiveness (MOEs) (e.g., driver and pedestrian behavioral data), driver mobility MOEs (e.g., travel times and speeds along corridors), and pedestrian mobility MOEs (e.g., pedestrian delays).

This report brings together the findings from the self-evaluations and contains cross-cutting analyses, where possible, of those countermeasures that were deployed by more than one of the field teams. Lessons learned by the field teams throughout the course of the project are also synthesized and presented herein.

RESULTS

This report presents and discusses the evaluation results for 18 pedestrian safety countermeasures (or combination of countermeasures). Ten of the 18 countermeasures were deployed by more than one of the field teams, and the remaining seven countermeasures were deployed by only one of the three field teams. For the purposes of presenting and discussing the results, the countermeasures were grouped into the following six categories:

• Static signs
• Active signs
• Pavement markings
• Signals and signal timing
• Physical separation
• Lighting

The findings are fairly mixed and in some cases inconsistent; however, this is not surprising considering the wide range of countermeasures installed, the various pedestrian safety problems at hand, the diverse locations and study sites at which the countermeasures were installed, and the somewhat different approaches to data collection and evaluation used by the three field teams. These were studies conducted in the field with real-world variables that cannot be controlled. Nonetheless, there were many notable and promising findings from the field tests and evaluations. A summary of the findings is as follows:

STATIC SIGNS

• TURNING TRAFFIC YIELD TO PEDESTRIANS signs. Installed at eight sites across the three field test locations, driver yielding behavior was the primary MOE for assessing the effectiveness of these signs. While there were a few significant changes found across the eight sites, there were inconsistencies in what changes were found and at which sites. These findings limit the conclusions that can be made regarding the effectiveness of these signs.
• In-street pedestrian signs. Installed at nine sites across the three field deployment locations, in-street pedestrian crossings signs appear to be highly effective at increasing driver yielding to pedestrians. The location at the roadway centerline appears to capture drivers’ attention more effectively than roadside signs, as evidenced by large increases in driver yielding at all but one of the nine sites. However, all three study teams noted that while these signs were effective at changing driver behaviors, they had a very short lifespan at the many of the sites. These issues can be overcome in a number of ways, including:
  • Placing the signs on raised medians as opposed to at street level
  • Placing only one sign at the crosswalk as opposed to using multiple signs on the approach
  • Avoiding use of the signs in locations with high truck or bus traffic
  • Carefully considering turning movements and lane width when determining locations for sign installation
• Pedestrian zone signs. Installed at one site in Miami, the results indicate that the countermeasure was not effective in reducing speed or increasing driver yielding / braking in the presence of pedestrians. The researchers have suggested that this ineffectiveness may be related to the low speeds observed prior to deployment, and therefore there was not much margin for improvement.

ACTIVE SIGNS

• NO TURN ON RED (NTOR) sign. Tested in Miami, and compared with both the static NTOR and the static conditional NTOR, the effectiveness of the electronic NTOR sign was assessed by observing driver violations of the NTOR restriction, right-turn drivers making complete stops, and pedestrian-vehicle conflicts. Use of the electronic NTOR sign resulted in the fewest turning violations (32 percent overall) of the three signs tested as well as the highest percentage of those violators that made a complete stop before violating the turn restriction. This sign may be especially effective in visually cluttered areas where motorists are less likely to see and respond to a static sign.
• Portable speed trailers. Installed in all three field locations, the primary MOE for assessing the effectiveness of speed trailers was average vehicle speed and driver yielding. The San Francisco team measured significant reductions in speed at their two test sites, while the Miami team did not. Significant increases in driver yielding at the San Francisco sites translated into decreases in pedestrian delays. There was an increase in driver braking mid-block in Miami, and no significant increase in driver yielding in Las Vegas. Based on these findings, it appears that the speed trailers can impact drivers' speeds and possibly increase their awareness of the presence of pedestrians at these locations.

PAVEMENT MARKINGS

• High visibility crosswalks. Tested at three locations in Las Vegas, there were no significant increases in driver yielding at any of the sites, and yielding distance results were inconsistent across the sites. There were significant reductions in drivers blocking the crosswalk at one of the sites. The results showed that high visibility crosswalks do not appear to be effective in changing driver behaviors in the vicinity of the crosswalks. This result could be due in part to the fact that the crosswalk markings deteriorated in a matter of weeks as a result of the heat causing a release of oils in the pavement.
• Advance stop lines. Installed at two locations in San Francisco, there were no significant changes in driver yielding, vehicle stop position, or pedestrian-vehicle conflicts at either site after installation of the advance stop lines. Based on these results, it appears that advance stop lines had no impacts on driver behavior or pedestrian safety.
• LOOK pavement stencils. Installed at four sites in San Francisco, there were few impacts on pedestrian looking behaviors and no impact on pedestrian-vehicle conflicts. Although the "LOOK" stencil markings are one of the least expensive countermeasures tested, the results indicate that this is not an effective countermeasure. Additionally, the San Francisco team noted that they were highly susceptible to fading and blemishes (similar to the high visibility crosswalk treatments in Las Vegas).

SIGNALS AND SIGNAL TIMING

• Pedestrian countdown signals. The findings from the Miami sites strongly point to overall increases in safe pedestrian behavior as a result of the pedestrian countdown signals, with significant and consistent positive results for all three critical MOEs: call button pressing, pedestrians in the crosswalk at the end of flashing DON'T WALK, and pedestrian signal violations. The results from the Las Vegas study team, however, were mixed, possibly due to signal timing issues at the intersections. The Las Vegas team also found a large increase in the percent of pedestrians that looked before crossing the street, which may have resulted from the animated eyes display on the countdown signals installed in Las Vegas.
• Call buttons that confirm the press. Installed in both Miami and Las Vegas at a total of three intersections, call buttons that confirm the press show a fairly strong and consistent positive impact on pedestrian safety in terms of increasing use of the call buttons and, in turn, reducing pedestrian violations and pedestrians trapped in the roadway. Call button presses increased significantly and to above 50 percent at both Miami sites, and pedestrian signal violations decreased at all three sites (however, overall pedestrian signal violations remained above 50 percent at both Miami sites). It was, however, difficult to see the LED light in bright Florida sunlight, making the auditory feedback more critical to the efficacy of the device at the Miami sites.
• Automated pedestrian detection (to activate or extend pedestrian crossing phase). Installed in both San Francisco and Miami, the only significant finding was a 9 percent reduction in the percentage of pedestrians trapped in the roadway at the Miami site. While these results suggests that pedestrians may have been making safer crossings, there were no measurable impacts of the pedestrian detection systems on pedestrian clearance (those clearing before the end of the WALK or clearance phases) or conflicts with motor vehicles (which were generally low to begin with).
• Activated flashing beacons. There were some clear increases in pedestrian safety where the activated flashing beacons were installed in San Francisco. There was a significant increase in driver yielding, corresponding decreases in pedestrian delay, and decreases in conflicts at both sites. There was an increase in yielding distance at one of the intersections in San Francisco and a decrease in pedestrians trapped at the other intersection. At the Las Vegas site, driver yielding did not change significantly, but this could have been a result of driver yielding improvements due to the installation of other countermeasures in earlier stages. For those drivers who yielded, yielding distances increased.
• Rectangular Rapid Flashing Beacons. The results of the study showed clear safety benefits associated with the introduction of the pedestrian activated RRFB in Miami. After installation of the RRFBs, driver yielding to both staged pedestrians and local resident crossings increased at both deployment sites, the percentage of pedestrians trapped in the middle of the road decreased at one of the sites, and evasive conflicts decreased at both sites. At one of the sites, the number of conflicts decreased each time the RRFB treatment was introduced and increased each time it was removed. At the other site, the decrease in conflicts after the RRFB was introduced was maintained each time it was removed. This may have represented some type of learning effect on the part of motorists.
• Leading pedestrian interval. Installed at four sites in San Francisco and two sites in Miami, the findings indicate that the countermeasure was effective at increasing left-turn driver yielding to pedestrians in the crosswalk, although the magnitude of left-turn yielding was smaller in San Francisco than in Miami (likely because left-turn driver yielding was already very high in San Francisco and therefore there was less opportunity for improvement). This effect does not appear to apply to right-turn driver yielding possibly due to the high frequency of right-turners who do not stop at a red light before turning. The Miami team also measured significant increases in pedestrian call button pushes and the number of pedestrians that start to cross at the beginning of the cycle.
• Prohibition of permissive left turns. Installed at one site in Miami, the data indicate that this countermeasure may be an effective way to improve pedestrian safety at intersections by reducing pedestrian-vehicle conflicts; however, the findings also indicate that there was a substantial portion of left-turners that violated the red signal. While this countermeasure has potential for increasing pedestrian safety, the signal configuration should be taken into consideration in order to mitigate left-turners violating the signal.

PHYSICAL SEPARATION

• Median refuge islands. Based on the results, it appears that the installation of a median refuge island at a mid-block location was effective in increasing driver yielding to pedestrians and reducing pedestrian delay, while the median refuge islands at the signalized intersections in San Francisco appear to be less effective at altering driver and pedestrian behaviors.
• Danish offset (in combination with high-visibility crosswalk, advance yield markings, and YIELD HERE TO PEDESTRIANS signs). Installed at two sites in Las Vegas (at one mid-block location and at one signalized intersection), this combination of countermeasures appears to have led to an increase in safe pedestrian and driver behaviors. The Las Vegas team measured significant increases in driver yielding and diverted pedestrians as well as significant decreases in trapped pedestrians. Pedestrian delay was significantly reduced at the mid-block location where a designated crossing area had not previously existed, although pedestrian delay increased at signalized intersection. There was no significant impact on vehicle delay at Lake Mead even though there was an increase in yielding. While driver yielding did increase significantly at the two locations, only 40 percent of drivers yielded at the mid-block location after installation of the countermeasures, while 76 percent of drivers yielded at the signalized intersection after installation of the countermeasures. This could be a result of the location of the Danish offset, the type of Danish offset that was installed, and/or whether or not a crosswalk existed in the baseline condition. At the signalized intersection site, the Danish offset was made more visible with the use of bright yellow bollards and there was a crosswalk in the baseline condition. At the mid-block location, the Danish offset was perhaps less visible and was located where there was not previously a crosswalk. In addition, vehicle speed could also play a role in the results. At the midblock location, there was a posted speed limit of 45 mph, while the posted speed limit through the signalized intersection was 30 mph. Drivers may be more willing and able to yield on the lower speed roadway. In general, though, the suite of countermeasures appears to have made pedestrian crossings safer.

LIGHTING

• Dynamic lighting. The findings from the Las Vegas team that tested the impacts of dynamic lighting at a high-visibility crosswalk location suggest that dynamic lighting used with automatic pedestrian detection increases pedestrian safety. Driver yielding and pedestrian diversion increased significantly while the percent of trapped pedestrians significantly decreased. While driver yielding increased, its prevalence was still low at 35 percent. In Miami, the addition of dynamic lighting to a crosswalk that had a rectangular rapid flashing beacon did not appear to further improve driver yielding or pedestrian-vehicle conflicts. The Miami researchers suggested that this may have occurred because the dynamic lighting is not very noticeable in the presence of the highly intense flashing beacons.

LESSONS LEARNED

Implementation and evaluation of the Pedestrian Safety Engineering and ITS-Based Countermeasures Program was challenging. The major steps in the project included:

• Establishing and maintaining a multi-agency pedestrian safety team to oversee and guide the project
• Identifying pedestrian safety and mobility problems, including potential contributing factors to crashes
• Selecting pedestrian safety countermeasures corresponding to the problems identified
• Obtaining funding and support for pedestrian safety improvements
• Procuring, deploying, and maintaining the countermeasures
• Evaluating the effectiveness of the countermeasures

Each step of the project offered new challenges to the project partners that are presented here as lessons learned. The lessons learned include general lessons learned and countermeasure-specific lessons learned. General lessons learned include the following:

• Assemble a diverse set of project partners to address the range of issues that might arise during the study
• Implement regular communication and participation mechanisms for project partners from project kick-off
• Use a variety of methods/sources to understand problems and to determine causes of crashes at prominent pedestrian crash locations
• Begin the program by implementing low-cost countermeasures for the greatest potential of widespread use
• Pursue a variety of funding sources for the pedestrian safety program
• Do not underestimate the complexity of procurement
• Budget ample time for deployment and coordinate with the appropriate jurisdictions
• Consider how the timing of countermeasure deployment may impact the experimental design and evaluation
• Consider the unique aspects of collecting and reducing pedestrian safety data

Countermeasure-specific lessons learned include the following:

• Strategically place in-street pedestrian signs to reduce the chance of them being hit by vehicles and to maximize their effectiveness
• Consider the technical issues surrounding the use of automated pedestrian detection
• Translate public service messages into multiple languages in order to conduct a successful outreach to non-English speaking populations
• Be prepared to demonstrate to concerned traffic engineers that the electronic NTOR sign will not significantly disrupt traffic progression along a corridor. Work with the local electrical department and vendors to make sure everything is in place for success.

CONCLUSIONS

Overall, the implementation and evaluation of a comprehensive pedestrian safety program proved to be a very challenging undertaking for each of the three field teams involved. There were many lessons learned over the course of the 6-year project, ranging from assembling and maintaining communications with a diverse set of project partners, to countermeasure selection and procurement, to the details associated with the successful application of particular countermeasures.

The various countermeasures were classified according to effectiveness using overall findings of the field studies: high effectiveness, moderate effectiveness, low effectiveness, and effectiveness depends on application. While based on the field teams' results, the classification was subjective in nature. However, the classification of the countermeasures in this way has been done in an attempt to give the reader an idea as to which countermeasures may have the most promise in impacting pedestrian safety and which others may not.

Five of the countermeasures were classified as being highly effective in impacting behaviors related to pedestrian safety. These five countermeasures cover a range of applications, including signal timing, active and in-street signs, call buttons that provide feedback, and roadway design elements. Each of the countermeasures offers something unique over traditional countermeasures, whether it be providing additional information to pedestrians, high visibility to pedestrians or motorists, or an advantage to pedestrians when crossing. Therefore, it is not surprising that these countermeasures resulted in the most positive impacts. They include:

• Leading pedestrian interval
• Pedestrian countdown signals
• In-street pedestrian signs
• Call buttons that confirm the press
• Danish offset combined with high-visibility crosswalk, advance yield markings, and YIELD HERE TO PEDESTRIANS signs

Four of the countermeasures were classified as being moderately effective in impacting behaviors related to pedestrian safety. These countermeasures were the most difficult to classify in that there were positive findings, yet the findings were either mixed or inconsistent either within or across the field locations. In addition, they are all considered as "active" countermeasures that might have potential to increase safety. They include:

• Activated flashing beacons
• Electronic No Turn on Red (NTOR) sign
• Prohibition of permissive left turns
• Portable speed trailers

Five of the countermeasures were classified as having low effectiveness in impacting behaviors related to pedestrian safety. Three of these countermeasures were pavement markings and two of the countermeasures were static signs. These five countermeasures are static and it is not surprising that they did not produce more significant results when compared against the active and more innovative devices. The low effectiveness countermeasures include:

• High visibility crosswalks
• Advance yield markings
• LOOK pavement stencils
• TURNING TRAFFIC YIELD TO PEDESTRIANS signs
• Pedestrian zone signs

The effectiveness of three of the countermeasures seemed to depend mostly on the application, with positive impacts in one application and less positive impacts in another application. These countermeasures include:

• Automated pedestrian detection (to activate or extend pedestrian crossing phase)
• Median refuge island
• Dynamic lighting