Pedestrian Safety

Overview

In 2018 there were 6,283 pedestrians who died and approximately 75,000 were injured in traffic crashes in the United States (NCSA, 2020). Pedestrians accounted for 17% of total traffic fatalities and 3% of total people injured. Since 2003 there has been a gradual rise in the proportion of total pedestrian fatalities. Of the pedestrian fatalities in 2018 (NHTSA - CRSS, 2018; NHTSA - FARS, 2018; NCSA, 2020):

• 69% of pedestrians killed were males;
• 33% of pedestrians killed had BACs of .08 g/dL or greater;
• 81% of pedestrian fatalities occurred in urban areas;
• 74% of pedestrian fatalities occurred at non-intersection locations;
• 76% of pedestrians were killed in collisions that occurred when it was dark;
• The average age of a pedestrian killed was 48; injured was 39;
• Of all motor vehicle crash fatalities among adults, 21% of those 50-54, 21% of those 55-59, 22% of those 60-64, and 20% of those 65-69 were pedestrians. Adults 55-59 also had the highest number of pedestrian fatalities (608).
• The pedestrian fatality rate among children and teens (range of 0.29-1.08 fatalities per 100,000 population for ages 0-19) was lower than the pedestrian fatality rate for all adult age groups (range of 1.97-2.85 per 100,000 population).
• Child pedestrians 14 and younger accounted for 17% of the child motor vehicle fatalities and an estimated 11% of all pedestrians injured in traffic crashes;
• Of all the adults 65 and older killed in motor vehicle crashes, 18% were pedestrians, including 361 pedestrians 80 years and older; and
• The highest total pedestrian injury rates by age group were those 15 to 19 and 20 to 24 (31 and 32 per 100,000 population, respectively).

Crash Trends. Until the last 5 years, average pedestrian fatalities per year had dropped over the past 20 years, from an average of about 5,600 annually from 1991-1995 to an average of about 4,600 annually from 2009-2014. However, the percentage of pedestrian fatalities started increasing in 2009, leading to a new trend where the number of pedestrians killed averaged 5,768 annually during the 5 most recent years (2014-2018, see figure below). In 2018 the number of pedestrians killed was 6,283, up 3.4% from 6,075 in 2017 (NCSA, 2020), which was the highest total since 1990. Fatality rate trends—or fatalities adjusted per number of walking trips or miles traveled by walking—are unavailable because there is no systematically collected and consistent measure of walking (exposure) to estimate and compare fatality rates each year. The population-based fatality rate for pedestrians overall was 1.92 per 100,000 population, with a rate of 2.71 for males and 1.14 for females (NCSA, 2020). Males 80+ years had the highest fatality rate (4.33) among all age and gender groups. However, population-based rates do not fully account for trends in amounts of walking.

                        Source: NCSA, 2020

The average age of pedestrians killed has remained similar over the past 10 years, increasing from 45 to 48. As shown in the figure below, numbers of pedestrian fatalities among age groups fluctuated over the last decade. The largest decrease occurred in the age group 10 to 15, and the largest increase occurred in the age group 55 to 64. Not all fluctuations in pedestrian fatalities are explained by changes in population by age group, as the population-based rates have also varied by year. (Fatalities and fatality rates by population for age groups are available for 2018 in NHTSA’s Traffic Safety Facts report, NCSA, 2020).

Comparing average fatalities for the range 2014-2018 to those from 2009-2013, pedestrian fatalities decreased for the youngest age groups - less than 5, 5 to 9, 10 to 15, while all older age groups saw increases in fatalities (see figure below). For those 55 to 64, a notable increase in pedestrian fatalities was observed from around 674 per year to around 1,046 per year. As noted earlier, this increase may reflect increases in population and walking among this age group; however, not all the changes are explained by changes in population by age group, as the population-based rates have also varied by year. (Fatalities and fatality rates by population for  age groups are available for 2018 in NHTSA’s Traffic Safety Facts report, NCSA, 2020). Older pedestrians (generally those over 65) are more likely to die from their injuries when struck due to the inherent fragility associated with the aging process. Factors that may increase vulnerability to being struck for some older pedestrians include age-related physical changes that may lead to walking more slowly; difficulty crossing the curb, difficulty judging walking speeds and oncoming vehicle speeds, and difficulty with interactions dealing with turning vehicles at intersections; and possible confusion about pedestrian signal phases (Dommes et al., 2012; Holland & Hill, 2010; Coffin & Morrall, 1995).

Source: FARS data (NHTSA, 2018)

Note that different age group spans are used. The intent of the chart is to compare general trends in different age groups, not to compare fatalities by age.

 

Several studies have also noted the overrepresentation of minorities, immigrants, and low-income populations in pedestrian-vehicle crashes (Anderson et al., 2010; Chakravarthy et al., 2012; Chen et al., 2011; Lin et al.; Murtha, 2005); however, the causes and contributing factors of these elevated crash rates are not well understood. Some studies attribute higher minority crashes to potential inequities in how pedestrian facilities are distributed across areas with different socioeconomic indicators (Kravetz & Noland, 2012). Other studies have found that lower income and minority populations have higher transit use and walking rates (or exposure) that may help partially explain elevated crash figures (Cottrill & Thakuriah, 2010). An analysis of pedestrian-involved traffic crashes in Florida found that among other factors, higher densities of discount grocery and convenience stores, banks, barber shops, and fast food outlets were associated with higher crash frequencies in areas with higher densities of low-income and minority populations (Lin et al., 2017). Still others have postulated that social-behavioral mechanisms and differing “safety cultures” play a role in pedestrian crashes, particularly for recent immigrants (Chen et al., 2011).

Despite the vulnerability of these groups to pedestrian crashes, the effectiveness of countermeasures in reaching these special populations is both unknown and challenging to evaluate. This is due to the lack of information about pedestrian safety programs targeted to the specific needs of low-income, minority, or immigrant populations, and because the courses or programs targeting these groups have been unsuccessful in measuring changes in behaviors. NHTSA developed and pilot-tested two English as a Second Language (ESL) courses to teach basic walking and bicycling safety concepts to adult immigrants learning the English language (see nhtsa.gov/pedestrian-safety/english-second-language-esl-teachers-and-learners). Both courses are free for use by formal programs or less formal settings with volunteer instructors. While NHTSA was able to evaluate the ease of use of these courses by ESL instructors and found increases in pre/post knowledge for the beginning level course, it was unable to successfully evaluate behavior changes as a result of this knowledge. A separate resource, the Resident’s Guide for Creating Safer Communities for Walking and Biking, demonstrated for use in communities, generated several case studies on inclusive approaches to outreach, community-based planning, and improving conditions for pedestrians, and was part of the update to the guide (Sandt, Thomas, et al., 2015). Finally, another challenge noted by NHTSA relates to the translation of educational material. Among some non-English speakers living in the United States, translating material is ineffective because they are not literate in their native languages. This knowledge spurred NHTSA to develop visual educational tools (motion graphics) to teach desired pedestrian behaviors, bicyclist behaviors around motorized traffic, and safe driver behavior around pedestrians and bicyclists. The motion graphics demonstrate motions visually without language to deliver safety education and are specifically designed for audiences that lack English language skills or literacy in their native languages as well as visual learners.

Walking Trends. Walking trends can be used to estimate exposure. The National Household Travel Survey (NHTS), conducted by the FHWA, captures walking and other travel trends in the United States. According to estimates from these surveys, the number of walking trips changed from 20.3 billion in 1995 to 33.1 billion in 2001, and to about 41 billion in 2009 (Santos et al., 2011). It is likely that at least some of the increases in 2001 and 2009 relate to more detailed questions prompting respondents to include walk trips in those 2 years, which was not done in the prior surveys. The latest 2017 NHTS was conducted from April 2016 to April 2017 by McGuckin and Fucci (2018). Major changes were made to the survey methodology including the use of address-based sampling, a web-based survey instead of a telephone survey, and the use of Google API to calculate trip length (see McGuckin & Fucci, 2018, and nhts.ornl.gov/ for more details). During the 2016-2017 survey period, an estimated 39 billion walking trips were made for all purposes (work or work-related commute, shopping and errands, school and church, social and recreational, and other trips). The 39 billion walking trips in 2017 represent approximately 10.5% of all transportation mode trips reported. About 4% of all trips to work were made by walking. Commuting to work, however, makes up only a small percentage (6.5%) of all walking trips. The largest proportion of walking trips were made for recreational and social reasons (47.5%) followed by shopping and personal errands (29.5%). McGuckin and Fucci found walking to school or church made up nearly 10.7% of walking trips (Table 9a). The percentage of students walking to school has also increased from 11.9% in 2007 to 15.2% in 2014 for morning trips and from 15.2% to 18.4% for afternoon trips (National Center for Safe Routes To School, 2016). This represents increases of 32% and 24%.

The increase in walking trips is especially significant since it represents increases in the average number of daily walking trips per person (Pucher et al., 2011), whereas total daily personal trips per person have been declining since the 1995 survey (Santos et al., 2011). The CDC’s National Health Interview Survey, collected in 2005 and 2010, assessed changes in prevalence of walking for at least 10 minutes one or more times in the preceding 7 days. Walking prevalence increased significantly, from 55.7% in 2005 to 62.0% in 2010. In 2010 walkers were also significantly more likely to meet the aerobic physical activity guidelines (CDC, 2012). The 2015 National Health Interview Survey found that 63.9% of adults reported walking for at least one bout of 10 or more minutes in the preceding week (Ussery et al., 2017; NPAP, 2017). There were increases in the prevalence of walking among adults between 2005 and 2010; however, the increase plateaued from 2010 to 2015 among adults (Ussery et al., 2017). The CDC encourages walking and bicycling to help meet physical activity guidelines. The CDC also supports building communities that provide safe and equitable opportunities to walk such as implementing Complete Streets policies and designs and lowering speed limits in urban areas, etc. For more information see www.cdc.gov/nccdphp/dch/programs/communitiesputtingprevention­towork­/resources/physical_activity.htm. Also see Health Resources in Action’s web pages on Community Speed Reduction and Public Health, www.hria.org/resources/reports/community-speed-reduction/2013-resources-speed-reduction.html. In 2015 the Surgeon Generalreleased Step It Up! The Surgeon General’s Call to Action to Promote Walking and Walkable Communities. Visit www.surgeongeneral.gov/library/calls/walking-and-walkable-communities/call-to-action-walking-and-walkable-communites.pdf (see also CDC, 2017).

Classifying Crash Types. Beginning in the 1970s, pedestrian crashes were categorized into types based on pedestrian and motor vehicle pre-crash actions, and crash location. In the early 1990s, this methodology was used to categorize more than 5,000 pedestrian crashes in California, Florida, Maryland, Minnesota, North Carolina, and Utah and analyze related characteristics (Hunter et al., 1996).

Of these 5,000+ pedestrian-motor vehicle crashes:

• 32% occurred at or within 50 feet of aintersections. Of these intersection crashes:

• 30% involved turning vehicles;
• 22% involved pedestrians dashing into intersections;
• 16% involved driver violations (e.g., running a red light);
• Older pedestrians were overrepresented in collisions with turning vehicles and motorist violations;
• Children were overrepresented in intersection dashes;

• 26% occurred at the middles of blocks (mid-block). Of these mid-block crashes:
  • 35% involved pedestrians running into the street and the drivers’ view was not obscured.
  • 17% were “dart-outs” in which pedestrians walked or ran into the street from locations where the pedestrians could not be seen.
  • Children were also overrepresented in dash-and-dart-out crashes;
• 7% occurred walking along roadways, not on sidewalks. Of these crashes:
  
  • 73% of the pedestrians were struck from behind while walking in the same direction as traffic;
  • Darkness and rural locations were overrepresented. This association is expected since rural areas are less likely to have sidewalks and supplemental street lighting.

The Pedestrian and Bicycle Crash Analysis Tool (PBCAT) software helps jurisdictions type pedestrian crashes to develop a database for analyzing pedestrian crash problems. Crash typing methodology has been used to develop tools that communities or States may use to discover more information about pedestrian and bicycle crashes. They can use crash type information and other crash characteristics to help select appropriate countermeasures. It is important to consider on-site field review of behaviors and site-specific characteristics before determining whether specific enforcement, educational, or engineering countermeasures are appropriate (Zegeer et al., 2008). Research has begun disentangling the behavioral elements included in some of the PBCAT crash types (Schneider & Stefanich, 2016). The study found that pedestrian crashes were more prevalent on the far side than the nearside of intersections, and were more likely fatal when the impacting vehicle had been traveling straight, if they were on roads in between intersections, and if they involved pedestrians approaching from the left of the vehicle. PBCAT may be downloaded from www.pedbikeinfo.org/pbcat_us/. Registration is requested for this free software, so the user may receive software updates or important technical information.

Underreporting and Crash Analysis. Another consideration when analyzing crash data is that pedestrian as well as bicycle crashes tend to be underreported. Many States may not require reporting nor collect off-road or private-road crash records. Non-roadway crashes may, however, constitute a significant portion of pedestrian-related crashes with motor vehicles. In several studies, parking lot and driveway-related crashes represented up to 15% to 25% or more of all reported pedestrian crashes (Stutts & Hunter, 1999a; Thomas & Levitt, 2014). Many more roadway and non-roadway crashes go unreported. Research is needed to better understand the extent and causes of non-roadway pedestrian crashes and effective countermeasures. NHTSA’s Not in Traffic Surveillance (NiTS) monitors and reports on not-in-traffic-related motor vehicle deaths. Many events involve young children. See Section 1.1 for more information.

Underreporting of traffic-related crashes on road right of ways likely decreases as the crash severity increases because police are likely to be called to injury and fatal crashes, and the pedestrian is more likely to be transported or seek examination at a healthcare facility. Therefore, the FARS data presented earlier are thought to be reliable sources for estimating pedestrian fatal crash frequencies. Even so, not all fatal pedestrian crashes are included in FARS, including fatal pedestrian crashes involving bicycles, or those that did not occur on public roads, as already mentioned.

Many more pedestrian and bicyclist injuries, including those due to falls, collisions with bicycles, and others, likely go unreported to State crash databases (Stutts & Hunter, 1999a, 1999b; Sciortino et al., 2005). A study of crashes reported to police and in emergency rooms from 2003-2007 in Funen, Denmark, estimated that 44% of severe and 76% of slight pedestrian injury crashes were not captured in police reports (Janstrup et al., 2016). Injured pedestrians may only report to the emergency rooms when coming in for treatment without informing the police of the crashes (Janstrup et al., 2016); this shows the need for analyzing emergency room records in addition to police records when developing injury estimates. Research is also needed to better understand the causes of these types of injuries. Maintenance of surfaces and Americans With Disabilities Act-compliant design of sidewalks, landings, and access ramps are certainly important for maintaining smooth surfaces and safe and accessible sidewalks and ramps. Other measures, such as providing space for bicyclists to ride separated from pedestrian walkways, may also be important but are outside the scope of this document.

Pedestrian Attributes – Everyone is a pedestrian, though when asked a person may not think of being one until prompted. Pedestrians span the full spectrum of ages from babies pushed in strollers to older adults. This includes foreign visitors and immigrants used to different traffic conventions, who speak many languages, and who may not be literate in their native languages. Pedestrians include disabled people who may be visually impaired, hearing impaired or deaf, or may require devices like walkers, wheelchairs, or crutches. More generally, we are all pedestrians when we walk our dogs, cross the street to talk to neighbors, go to the store, go to school, or walk to or from the bus stop.

Crash Factors – A large body of historical research has established numerous factors associated with pedestrian crashes. Pedestrian and driver pre-crash actions and behaviors (such as distraction, driver speed, and alcohol use, and vehicle type and design) all contribute to pedestrian crashes. Crash factors are not mutually exclusive and in many ways, have compounding effects on pedestrian risk. More details about these factors are discussed below. Several studies have provided evidence of the role of the transportation environment in pedestrian safety and summarized best practices in engineering and design for pedestrian safety (FHWA, 2010; Redmon, 2011; Retting et al., 2003). Complete Streets (also known as Livable Streets) policies are one of the more low-cost and effective countermeasures, as evidenced by numerous cities and States across the United States (FHWA, 2010). For more on Complete Streets, visit www.smartgrowthamerica.org/complete-streets/. Also, search for a program in your State or city. Vision Zero is another large program that aims to reduce pedestrian fatalities through a focus on engineering changes and speed management (Kim et al., 2017).

Distraction – Cell phones and electronic devices are sources of distraction, not only for motorists (discussed in the Distracted Driving chapter), but also for pedestrians. A literature review from NHTSA found that, based on the limited amount of research done on pedestrian distraction, distraction is associated with a small but statistically significant decrease in pedestrian safety (Scopatz & Zhou, 2016). Talking on cell phones is associated with cognitive distraction that may reduce the frequency of prudent pedestrian behaviors, particularly among college-age pedestrians who may be more engaged with such devices (Hatfield & Murphy, 2007; Nasar et al., 2008; Ortiz et al., 2017; Stavrinos et al., 2009, 2011); however, the results from real-world observational studies are mixed (Walker et al., 2012; Thompson et al., 2013). A study of road user distraction at four intersections in Washington, DC, found that of the 4,871 people observed, the primary form of distraction was engaging with other people for both pedestrians (44%) and drivers (49%) (Ortiz et al., 2017). More than a quarter (27%) of all people observed were distracted by cell phone use. The prevalence of distraction among pedestrians was higher than for drivers: pedestrians had 1.5 increased odds of being distracted. Thompson et al. (2013) sampled pedestrian behaviors at 20 high-risk intersections and reported that only pedestrians who were texting were associated with suboptimal crossing behaviors. Ortiz et al. (2017) reported that the majority of interactions (20 of 21 or 95%) between distracted pedestrians and distracted drivers resulted in some form of evasive maneuver by either.

These studies report 7% to 30% of pedestrians using varyious portable electronic devices. Nationally representative estimates on use of portable electronic devices are unavailable, but would likely only capture a snapshot in time, as device use continues to grow in popularity. FARS/GES data on pedestrian device use or involvement in pedestrian crashes are unavailable at the national level.

Driver speed – Driving speed is a key risk factor in severe pedestrian crashes. The study by Rosen and Sander (2009) is believed to be one of the more robust in terms of estimating the risk of pedestrian fatality based on driver impact speeds. The study estimated fatality risk curves based on driver impact speeds, ranging from 8% at 50 km/h (31 mph) and reaching 50% at 75 km/h (about 47 mph). Other studies have estimated similar relationships, although the magnitude varies (Leaf & Preusser, 1999; Tefft, 2011). As pedestrians are particularly vulnerable to severe injury and fatality when struck by higher-speed vehicles, countermeasures aimed at reducing vehicle speeds have the potential to save lives for both pedestrians and drivers. Driving speed also appears to affect the tendency for drivers to yield to pedestrians at crosswalks, with fewer drivers yielding as speeds increase (Bertulis & Dulaski, 2014; Gårder, 2004). Speeding-related countermeasures are presented in the Speeding and Speed Management chapter.

Risk of severe injury (left) and death (right) in relation to impact speed in a sample of 422 pedestrians  15+ years struck by a single forward-moving car or light truck model year 1989–1999, United States, 1994–1998. Risks are adjusted for pedestrian age, height, weight, body mass index, and type of striking vehicle, and standardized to the distribution of pedestrian age and type of striking vehicle for pedestrians struck in the United States in years 2007–2009. Dotted lines represent pointwise 95% confidence intervals. Serious injury is defined as AIS score of 4 or greater and includes death irrespective of AIS score.

Source: Tefft (2011)

Alcohol – The role of alcohol in pedestrian crashes has not been well defined, based on the lack of complete and high-quality data on alcohol use or BACs of drivers and pedestrians involved in crashes. Driver or pedestrian alcohol use is estimated to be a contributing factor in 48% of pedestrian fatalities (NCSA, 2020). Thirty-three percent of pedestrians killed in crashes had BACs of .08 or higher, while 16% of fatal pedestrian crashes had drivers with BACs of .08 or higher. From 1982 to 2014 the proportion of fatally injured pedestrians with BACs of .08 or higher decreased at a lesser rate than fatally injured passenger drivers with BACs of .08 or higher (Eichelberger et al., 2018). One recent study of alcohol outlet density and pedestrian safety in Baltimore found that each additional off-premise alcohol outlet was associated with a 12.3% increase in the risk of pedestrian injuries in the neighborhood, and an attributable risk of 4.9% (Nesoff et al., 2018). Alcohol-related countermeasures that may help address certain pedestrian crashes are presented in the Alcohol- and Drug-Impaired Driving chapter.

Vehicle Type and Design - Previous studies have focused on the role of vehicle type, design, and warning systems in the event of crashes (Searson & Anderson, 2011), and in the ability of pedestrians and even vehicle technology to detect and prevent crashes (Fredriksson et al., 2011; Greene et al., 2011). A study of pedestrian, bicyclist, and motorcyclist collisions with reversing cars conducted in Germany found that pedestrians were seriously injured in 9.1% of all such crashes from 1999 to 2012 (Decker et al., 2016). Of the crashes for which impact zone data were available, impact with the rear of the car was the most frequent (81%), followed by impact with the side (18%). Backup cameras in vehicles can help prevent these crashes with reversing cars. See www.nhtsa.gov/equipment/driver-assistance-technologies for descriptions of in-vehicle driver assistance technologies. Another issue in the literature, as hybrid and electric vehicles constitute larger portions of the vehicle fleet, is the consequence of “quiet” vehicles on pedestrian safety, particularly among pedestrians with visual disabilities who rely more on auditory cues to detect traffic (Garay-Vega et al., 2011).

Engineering and Roadway Design. While not dismissing the importance of vehicle design and the role of the built environment in preventing pedestrian crashes, the countermeasures described in this guide relate primarily to educational and enforcement measures aimed at improving the knowledge and behaviors of road users to prevent crashes. However, there is a growing recognition of the importance of road design and the built environment in fostering safer user behaviors. A comprehensive approach that uses a combination of effective engineering, enforcement, and educational measures may have the best chance of achieving desired crash reductions. U.S. DOT released a national pedestrian safety action plan summary focusing significant attention on the built environment research and countermeasures (2014). Key infrastructure resources are included in the Resources section.

Emerging Technologies or Emerging Vehicle Technologies. Further, emerging research is exploring whether vehicle technologies known as Pedestrian Crash Avoidance/Mitigation (PCAM) systems show promise in reducing motor vehicle-pedestrian crashes (Yanagisawa et al., 2014; Yanagisawa et al., 2017). Current testing is largely limited to a research environment involving light vehicles and measuring the systems' capabilities to detect a pedestrian in the road ahead. The systems may alert drivers, automatically brake, or take other measures to prevent crashes with pedestrians. An increasingly relevant issue is the emergence of connected and automated vehicles, and research is required to investigate their safety and equity impacts on pedestrians and walkability (Shay et al., 2018). A small-scale crowdsourced survey reported that study participants perceived higher safety for pedestrians with automated vehicles (Deb et al., 2017); however, Millard-Ball argues that pedestrians may adapt to safety benefits offered by automated vehicles and expect the same from interactions with non-automated vehicles (Millard-Ball, 2018).

Safety in Numbers. Finally, the idea that vulnerable road user safety may be improved by increasing the numbers of pedestrians and bicyclists is gaining traction and empirical support (Elvik & Bjørnskau, 2017; Elvik & Goel, 2019; Jacobsen et al., 2015). As numbers of pedestrians increase, drivers should expect to see more pedestrians and thus become more attentive to them. This effect has been found to be consistent when analyzing data from communities of varying sizes and population scales (Jacobsen et al., 2015). A 2009 scanning tour by U.S. transportation officials and researchers of Denmark, Sweden, Germany, Switzerland, and the United Kingdom reported that the concept of “safety in numbers” has motivated the promotion of more bicycling and walking in these countries as a safety countermeasure (Fischer et al., 2010). However, encouragement in these countries is done in the context of commitments to comprehensive planning, funding, engineering, and design and maintenance policies to provide safe and connected pedestrian networks. The scan report also documents numerous examples of how these policies are put into practice through traffic calming, traffic and parking management, enforcement, education and other systemic approaches. Research from abroad as well as the United States finds that, although the actual number of crashes may go up with increases in walking (and bicycling) with increased exposure, individual risk of crashes with motor vehicles (crash rate) is lower as numbers of pedestrians and bicyclists increase (Alliance for Biking and Walking, 2014; Geyer et al., 2006; Jacobsen, 2003: Jacobsen et al., 2015; Leden et al., 2000). The European countries mentioned above are also committed to reducing the total numbers of pedestrian fatalities and injuries while increasing walking and bicycling. Many European countries have adopted “toward zero deaths” safety philosophies (for more information see - https://ec.europa.eu/transport/road_safety/home_en). In the United States, the “Vision Zero” initiative primarily targets local jurisdictions to get them to adopt speed-management policies and roadway design practices that encourage driving at speeds that are less likely to result in serious injuries or fatalities. As of mid-2018, more than 30 cities had adopted policies from this initiative (Vision Zero Network, 2018).

A non-linear relationship between traffic volumes (motorist, pedestrian, or bicyclist) and crashes has long been demonstrated (AASHTO, 2010; Bhatia & Wier, 2011; Elvik & Bjørnskau, 2017), but a causal mechanism for how increased volumes improve pedestrian safety has not been demonstrated (Bhatia & Wier, 2011). This means that crashes do not tend to increase in direct proportion to increases in volume, but absolute crash numbers are still likely to increase (and have increased) with increases in walking – all else being equal. Additionally, all the studies cited above, and others attempting to characterize pedestrian safety relationships, are based on cross-sectional comparisons. There are frequently safety factors such as motorist speed, congestion, or law enforcement activity that are unmeasured or have not been accounted for in such studies. However, a recent meta-analysis of motorist-pedestrian or motorist-bicyclist injury crashes accounted for some of the similarities among factors included in previous studies and estimated that there is safety in numbers for both pedestrians and bicyclists (Elvik & Bjørnskau, 2017). By their estimate, if the number of pedestrians or cyclists doubles (100% increase), the increase in accidents is expected to be 41%, which is much less than direct proportionality. A subsequent expanded meta-analysis determined that the safety in numbers effect may be much stronger for pedestrians than cyclists, and at the macro level (cities) than the micro-level, such as an individual junction (Elvik & Goel, 2019).

One issue that makes the direct application of the safety in numbers findings difficult is that the cross-sectional studies cannot demonstrate the direction of effect – that is, whether a safer environment comes before the increased number of crashes or as a result (Bhatia & Wier, 2011). Impaired pedestrians also contribute to the overall safety problem; however, more research is needed about this issue in general, in addition to a better understanding of how laws and education can mitigate risks posed by impaired pedestrians. It is clear, however, that a focus on improving the environment, both infrastructure and road users’ compliance with laws and safe behaviors, is important to increasing both population-level safety (measured as a reduction in population-wide fatalities and injuries) and numbers of pedestrians or amounts of walking. As these two elements – safety improvements and increases in walking – go together, individual risk will also be reduced.

---