Enhancing Pedestrian Volume Estimation and Developing HCM Pedestrian Methodologies for Safe and Sustainable Communities (2022)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

168 Appendix A: Literature Review Introduction This appendix presents the literature review for NCHRP Project 17-87, “Enhancing Pedestrian Volume Estimation and Developing HCM Pedestrian Methodologies for Safe and Sustainable Communities,” summarizing information relevant to the project objectives from over 300 reviewed documents. Because members of the project team have performed literature reviews for other related projects in the recent past, the information obtained from the following reviews was used as a starting point for this review:  NCHRP Project 17-84, “Pedestrian and Bicycle Safety Performance Functions for the Highway Safety Manual”;  NCHRP Project 07-19, “Methods and Technologies for Collecting Pedestrian and Bicycle Volume Data”;  NCHRP Project 07-17, “Pedestrian and Bicycle Transportation Along Existing Roads”;  NCHRP Project 17-73, “Systemic Pedestrian Safety Analysis”;  NCHRP Project 15-63, “Guidance to Improve Pedestrian and Bicycle Safety at Intersections”;  NCHRP Project 03-120, “Assessing Interactions Between Access Management Treatments and Multimodal Users”;  NCHRP Project 07-25, “Guide for Pedestrian and Bicycle Safety at Alternative Intersections and Interchanges (A.I.I.)”;  Oregon DOT SPR 778, Safety Effectiveness of Pedestrian Crossing Enhancements;  Oregon DOT SPR 779, Risk Factors for Pedestrian and Bicycle Crashes; and  FHWA Report HPL-16-026, Exploring Pedestrian Counting Procedures. The Transport Research International Documentation database was used to identify more recent literature, as well as relevant literature not included in the reviews listed above. When appropriate, literature referenced in documents identified through this process has also been included in the review. The literature review is presented in three main parts. The first part describes literature relevant to the project objectives in the following subject areas:  Techniques for Efficient and Accurate Estimation of Pedestrian Volume and Exposure  Performance Measures for Evaluating Pedestrian Safety, Operations, Mobility, and Satisfaction  Pedestrian Safety Countermeasure Effects on Pedestrian Safety, Operations, and QOS The second part of the literature review describes the existing HCM pedestrian operations and QOS methods for streets and roadways. This section describes the evolution of the methods from the 1985 HCM (TRB 1985) to the current HCM 6th Edition (TRB 2016), the input factors used by the current HCM methods, the sensitivity of results to each input factor, the identified limitations of the methods, and (when available) descriptions of applications of the methods in the literature. The final part summarizes the key findings from the literature review. A list of all documents referenced in the review is provided at the end.

169 Techniques for Efficient and Accurate Estimation of Pedestrian Volume and Exposure Methods to Collect Pedestrian Volume Data Counting pedestrian volumes can be challenging because pedestrians travel without restrictions and often in groups. Hence, automatic detection of pedestrians can be difficult, especially if multiple people travel through the sensor simultaneously. The most commonly used methods to collect pedestrian volume data include manual counts in the field, manual counts from video, automated counts from video, passive or active infrared sensors, piezoelectric pads, and radio beams. Table A1 provides a summary of some of the major advantages and disadvantages of the data collection methods identified in the research literature (Greene-Roesel et al. 2008, FHWA 2011, Ryan et al. 2013, Nordback et al. 2016, Ryus et al. 2016). More details on each counting method can be found in FHWA 2011. Surveys conducted by FHWA (over 100 responses) and NCHRP Project 07-19 (269 responses) of practitioners in state transportation agencies, MPOs, and local governments revealed that pedestrian data are often collected for safety analyses, trend analyses, before-and-after studies, project prioritization, network analysis, funding requests, and project designs (FHWA 2011, Ryus et al. 2014a). These surveys also revealed that most respondents’ agencies used manual or manual video counts. When automated counters were used, responding agencies most often utilized passive infrared, closely followed by active infrared. While the FHWA survey only found a relatively small proportion of respondents using automatic counts from video, the NCHRP survey conducted about five years later found nearly as many respondents using automatic counts from video as using passive infrared counters. Only a small portion of responding agencies utilized automatic counts from piezoelectric pads, laser scanners, or thermal imaging cameras. Challenges with Collecting Pedestrian Volume Data The most commonly cited challenges of collecting pedestrian volume data are cost and staff time, followed by lack of technological tools and lack of organizational interest in, or defined need for, pedestrian count data (Ryus et al. 2014a). One of the concerns is that agencies are not fully aware of the trade-offs between labor cost of manual counts and installation and maintenance costs of automatic counters (Schneider et al. 2005). Manual counts (on-site or video) have high labor costs but low setup and maintenance costs, while other technologies such as piezoelectric pads have high installation and maintenance costs. On the other hand, infrared sensors have relatively lower initial and maintenance costs than other technologies (FHWA 2011, Nordback et al. 2016, Ryan and Lindsey 2013, Ryus et al. 2014). Hence, these automated counters can often be cheaper in the long run than prolonged manual counting. Another concern is the accuracy of pedestrian counts. Manual on-site counts can have errors associated with observer fatigue of up to 15% and cannot be verified (Diogenes et al. 2007, Nordback et al. 2016, Ryus et al. 2014). Infrared sensors are typically viewed as the best automated counters for pedestrians, however, they tend to undercount pedestrians: studies have shown that infrared sensors count 9%–19% fewer pedestrians than those that pass the counter (Schneider et al. 2009, Greene-Roesel et al. 2008), but with substantial variations in accuracy between devices from different manufacturers (Ryus et al. 2016). Infrared sensors can have errors in detecting pedestrians if they travel next to each other (FHWA 2011, Nordback et al. 2016, Ryan and Lindsey 2013); active infrared sensors are also vulnerable to rain, leaves, etc. and are best suited for indoor use (FHWA 2011). Nevertheless, testing of many different counting devices found that reasonable error correction factors could be developed for all of the tested devices, including those with high over- or undercounting rates (Ryus et al. 2016).

170 Table A1. Methods for Counting Pedestrians. Counting Method Advantages Disadvantages Manual counts in the field  Can be accurate with count staff training  Can capture user characteristics and directional counts  Minimal equipment costs  High labor cost  Only for short-term counts  Data cannot be verified Manual counts from video  High accuracy (can pause recording and rewind as needed)  Can capture user characteristics and directional counts  Fewer personnel than in-field counts  Data can be verified  High labor cost  Low equipment cost Automatic counts from video  Good for crowded areas  Can be used for long-term counts  Some systems can track pedestrian and auto trajectories  Data can be verified  High equipment cost  High development cost  Accuracy not verified independently – different approaches Active infrared  Portable  Low equipment cost  Easy installation  Can be used for long-term counts  Undercounting due to occlusion (5%–15%)  Cannot distinguish bikes from pedestrians  Directional counts not possible  Requires emitter and receiver mounted on opposite sides of facility Passive infrared  Low equipment cost  Not affected by wet or foggy weather  Can be used for long-term counts  Widely used  Undercounting due to occlusion (5%–50%)  Directional counts not possible  Cannot distinguish bikes from pedestrians  Background objects (e.g., windows) may produce false detections Piezoelectric pads  Low maintenance cost  Low power consumption  Can count peds on sidewalks  High installation cost  May not be able to differentiate groups of pedestrians Radio beams  Low equipment cost  Portable  Some products can distinguish bikes from pedestrians  Undercounting due to occlusion (6%–12%)  Directional counts not possible  Requires emitter and receiver mounted on opposite sides of facility with relatively narrow separation  May be affected by metal objects (e.g., bridges with large metal components) Laser scanner  Easy setup  Large coverage area  Expensive  Performance impacted by weather conditions Thermal imaging camera  Can be used for long-term counts  Not affected by ambient light conditions  High installation cost

171 Counting Method Advantages Disadvantages  Can capture directional counts and potentially distinguish bicycles from pedestrians On shared-use facilities, more than one sensor type may be required to distinguish between pedestrians and bicycles—for example, a passive infrared sensor will detect all path users, while infrared loops can be used to detect only bicycles, with the pedestrian count being the difference in the two counts (Ryus et al. 2014). Some automated technologies, such as automated counts from video, thermal cameras, and radio beams may be able to distinguish bicycles and pedestrians using only one sensor. Some radio beam devices use multiple frequencies to distinguish pedestrians and bicycles, but may also have constraints on the maximum distance between emitter and receiver (e.g., 10–12 feet), which can constrain where they can be placed (Ryus et al. 2016). NCHRP Project 17-84 (Torbic et al. 2018) provides an extensive literature review on challenges related to collecting pedestrian volume data. A summary of this literature review is provided below. Table A2 summarizes the different barriers associated with each counting method. Temporal barriers refer to the limited duration of counts available (typically 2 hours) and lack of experience extrapolating short-term counts to longer time periods (Injury Surveillance Workgroup 8 2017, FHWA 2013, Ryus et al. 2014, FHWA 2011). Spatial barriers refer to the limited number of locations at which pedestrian counts can be conducted and the difficulty in selecting the correct locations to perform pedestrian counts (Nordback et al. 2016, Schneider et al. 2005). Legal barriers refer to obtaining permits for collecting pedestrian data and installing counting devices (Ryus et al. 2014). Documentation barriers refer to a lack of standard methods to collect pedestrian volume data and challenges associated with storing continuous data (Nordback et al. 2016, Griffin et al. 2014). Lastly, the funding barrier refers to the difficulty in obtaining budgets for conducting pedestrian volume counts where only three states have extensive support for pedestrian counts (Lindsey et al. 2013). Table A2. Barriers Associated with Pedestrian Counting Methods. Counting Method Barriers Temporal Spatial Legal Documentation Funding Manual counts in the field X X X X Manual counts from video X X X Automatic counts from video X X X Active infrared X X X Passive infrared X X X Piezoelectric pads X X X X Best Practices This section summarizes the literature on best practices for overcoming two main challenges associated with collecting pedestrian volume data: cost and accuracy. Additionally, best practices associated with the spatial and temporal barriers related to estimating exposure from volume data are summarized. Practices for Overcoming Cost, Staffing, and Funding Challenges Costs for pedestrian counting projects can be reduced and the challenges associated with limited staffing can be alleviated by:

172  Outsourcing labor by recruiting volunteers from universities or community groups or engaging different stakeholders (Lindsey et al. 2013). For example, Cheyenne, Wyoming, recruited local Boy Scouts to conduct over 40 pedestrian counts on the Cheyenne Greenway in 1996 and 1997 with no direct costs to the city government.  Crowdsourcing for labor, e.g., the Amazon Mechanical Turk website used by a study conducted at Washington University in St. Louis (Hipp et al. 2013). Methods to procure more funding for count programs also need to be considered. The most commonly used practice to increase funding for count programs is to justify the use of existing counts by tracking their usage and identifying purposes for their use such as active transportation or evaluating the need for improved facilities (Charlier Associates, Inc. et al. 2012; Schneider et al. 2005). Alternative funding sources have also been used to collect pedestrian volume data, e.g., Minnesota DOT has received grants from the State Health Improvement Program to conduct their counts as a part of active transport and healthy living initiatives. Practices for Overcoming Challenges Related to Data Accuracy Quality control is a method that can ensure a minimum level of accuracy for all pedestrian count data. Quality assurance methods have been implemented by NCDOT, where data are checked for potential equipment malfunction by checking for consecutive zeros and identifying data outside the standard range of data (Jackson et al. 2017). In another study, the interquartile range has been used to compare directional counts and to identify outliers (Turner and Lasley 2013). Practices for Estimating Exposure from Count Data Due to the challenges identified above, available pedestrian count data are often short-term and collected only at a few selected locations. Sometimes these short-term counts are expanded to weekly, monthly, or annual exposure using expansion factors, but they are still only applicable to the specific locations. Several methods have been developed for estimating pedestrian volumes at intersections and along roadway segments. NCHRP Report 770 (Kuzmyak et al. 2014) provides a summary of the existing pedestrian demand modeling research. In this work, three categories of models are described:  Trip generation and flow models. This approach estimates pedestrian trip generation based on land use and network connectivity and assign these trips to facilities. These models are similar to four-step transportation models, performing trip generation, distribution, and network assignment; however, they do not include a mode choice step. A pedestrian analysis zone is determined to generate the trips; however, trips assigned to the pedestrian network ultimately are totaled to obtain demand for intersections or road segments.  Network simulation models. This approach considers spatial characteristics and relationships, such as connectivity and sight lines, to simulate paths taken by pedestrians. In some cases, these network variables are combined with land use variables to estimate pedestrian volumes (Raford and Ragland 2004, Raford and Ragland 2005). These models often require coding of a detailed network and require proprietary software to apply. One commonly used software is Space Syntax, which was developed in London in the 1980s and is widely used in Europe for pedestrian planning.  Direct demand models. This is the most commonly used practice in the U.S. for estimating pedestrian demand. These models aim to explain pedestrian activity on road segments or at intersections using factors that are tailored to the local context. Commonly used model structures include linear, loglinear, or negative binomial regression. Variables used to represent the local context include population or employment density, land use, sidewalk presence, vehicular traffic interactions, transit availability, and school proximity. Developing these models require existing

173 pedestrian counts: for example, street block face or midblock count data have been used to model pedestrian volumes in New York (Pushkarev and Zupan 1972), Milwaukee (Benham and Patel 1977), and Minneapolis (Hankey et al. 2012, Hankey and Lindsey 2016). However, more recent direct demand pedestrian volume models have been developed from intersection crossing counts (Pulugurtha and Repaka 2008, Schneider et al. 2009a, Liu and Griswold 2009, Haynes and Andrzejewski 2010, Jones et al. 2010, Miranda-Moreno and Fernandes 2011, Schneider et al. 2012, Grembek et al. 2014). Direct demand modeling is used most commonly because the relationships between the site characteristics and generated demand are easy to follow and can be easily applied (Turner et al. 2018). These models are estimated by collecting pedestrian counts over short periods or time, which can be expanded to annual volume estimates (Ryus et al. 2014). These counts are used as dependent variables in the predictive models, and statistical software is used to identify the relationship between the pedestrian volumes and explanatory variables. Once the model is estimated, it can be used for estimating pedestrian volumes anywhere within the community. Performance Measures for Evaluating Pedestrian Safety, Operations, Mobility, and Satisfaction Performance Measures for Evaluating Safety A number of risk factors have been identified in the literature pertaining to pedestrian–vehicle crash frequency and severity. Various sources, including NCHRP Report 803 (Lagerwey et al. 2015), NCHRP Project 17-73 (Thomas et al. 2018), and a research report produced for the Oregon DOT (Monsere et al. 2017), were reviewed to identify these factors, which are summarized in Table A3. While this list is not exhaustive, it captures the key factors and their impacts on crash frequency and severity. A “+” symbol indicates a positive association of the risk factor with pedestrian crash frequency and/or severity, whereas a “−” symbol indicates a negative association. “+/+” indicates a positive association with crash frequency and severity. These factors have been categorized into roadway, intersection, traffic characteristics, land use, demographics and behavior, and weather and lighting. Key factors include vehicle volumes and speeds, pedestrian volumes, transit activity, and presence of specific land uses. NCHRP Project 07-17 (Lagerwey et al. 2015) developed a methodology to prioritize pedestrian and bicycle improvements along existing roads, called the ActiveTrans Priority Tool (APT). The methodology identifies nine factors important for the prioritization process: Stakeholder Input, Constraints, Opportunities, Safety, Existing Conditions, Demand, Connectivity, Equity, and Compliance. The safety factor in the APT methodology is evaluated in terms of reported pedestrian and bicycle crashes and crash rates. Safety-related variables that are suggested to be considered include: total pedestrian and bicycle crashes, fatal and severe injury pedestrian and bicycle crashes, proportion of pedestrians walking in the roadway, proportion of pedestrians complying with DON’T WALK signals, proportion of motorists complying with right-turn-on-red restrictions, proportion of motorists yielding to pedestrians in crosswalks, and number of near-misses involving pedestrians or bicyclists. Conditions that influence pedestrian or bicycle safety, comfort, and demand include traffic speed, traffic volume and composition (including proportion of heavy vehicles), right-turning traffic volume, right-turn-on-red restricted or allowed, signal timing (delay), traffic control type, crosswalk warning signage presence, sidewalk presence, number of travel lanes, number of right-turn lanes, outside through lane width, roadway pavement condition, total crossing distance, curb radius, median or crossing island presence, on-street parking presence and utilization, bicycle lane presence and width, paved shoulder presence and width, driveway frequency, presence and width of a buffer between sidewalk and moving traffic, traffic calming measure use, and sidewalk condition.

174 Table A3. Pedestrian Crash Risk Factors. Studies Categories Risk Factors Impact on Frequency or Severity Diogenes and Lindau 2010, Chimba et al. 2014, Palamara and Broughton 2013, Fitzpatrick et al. 2014, Garder 2004, Zegeer 2006 Roadway Road width + Thomas et al. 2017 Number of lanes crossed in one maneuver + Diogenes and Lindau, 2010 Sidewalk width + Schneider et al. 2004, Diogenes and Lindau 2010, Torbic et al. 2010, Miranda-Moreno et al. 2011, Pulugurtha et al. 2011 Bus stops + Zegeer et al. 2006, Lee et al. 2005 Undivided crossing + Palamara and Broughton 2013, Schneider et al. 2010, Zegeer et al. 2006 Median refuges − Wier et al. 2009, Fitzpatrick et al. 2014, Miranda- Moreno et al. 2011 Functional class (arterials) + Schneider et al. 2004, Miranda-Moreno et al. 2011 Street length + Schneider et al. 2001, Garder 2004, Diogenes et al. 2010, Palamara and Broughton 2013 Marked crosswalk − McMahon et al. 1999, Schneider et al. 2004 Sidewalks − Poch and Mannering 1996 Intersection Left-through lanes + Poch and Mannering 1996 Protected left-turn lanes + Schneider et al. 2010 Right-turn-only lanes + Martin 2006 Cycle length + Poch and Mannering 1996 Grade + Poch and Mannering 1996 Horizontal curve on opposing approach + Torbic et al. 2010 Number of lanes crossed in one maneuver + Poch and Mannering 1996, Garder 2004, Lee et al. 2005, Zegeer 2006, Basile et al. 2010 No traffic control +

175 Studies Categories Risk Factors Impact on Frequency or Severity Poch and Mannering 1996, Schneider et al. 2004, Martin 2006, Loukaitou et al. 2007, Zegeer et al. 2006, Wier et al. 2009, Schneider et al. 2010, Torbic et al. 2010, Pulugurtha et al. 2011, Miranda-Moreno et al. 2011, Palamara and Broughton 2013 Traffic characteristics Vehicular volume + Fitzpatrick et al. 2014 High heavy truck volumes +/+ Diogenes et al. 2010 Public transit volume + McMahon et al. 1999, Schneider et al. 2004, Pulugurtha et al. 2011, Harwood et al. 2008, Schneider et al. 2010, Zegeer et al., 2005 Pedestrian volume + Poch and Mannering 1996, McMahon et al. 1999, Garder 2004, Lee et al. 2005, Zegeer et al. 2006, Zegeer et al. 2006, Martin 2006, Chimba et al. 2014 Vehicle speed, speed limit +/+ Lee et al. 2005, Zegeer et al. 2006, Senserrick et al. 2014 Land use Urban areas + Lee et al. 2005, Zegeer et al. 2006, Senserrick et al. 2014 Rural areas −/+ Pulugurtha et al. 2011 Single family residential area − Pulugurtha et al. 2011 Urban residential– commercial area − Pulugurtha et al. 2011 Commercial center area − Pulugurtha et al. 2011 Neighborhood service district − Loukaitou et al. 2007 Percentage of Commercial area + Loukaitou et al. 2007 Percentage of high-density residential areas + Loukaitou et al. 2007 Percentage of vacant industrial and office land − Miranda-Moreno et al. 2011, Chimba et al. 2014 Presence of schools + Ewing et al. 2003 Block size − McMahon et al. 1999, Martin Allison 2006 Older neighborhoods +

176 Studies Categories Risk Factors Impact on Frequency or Severity Loukaitou et al. 2007, Wier et al. 2009 Demographics and behavior Population density + Ewing et al. 2003, Loukaitou et al. 2007, Wier et al. 2009, Miranda-Moreno et al. 2011 Employment density + Ewing et al. 2003, Martin Allison 2006, Torbic et al. 2010, Chimba et al. 2014, Thomas et al. 2017 Household income − Martin 2006, Chimba et al. 2014 Number of vehicles per household − Lee et al. 2005, Martin 2006, Zegeer et al. 2006, Diogenes et al. 2010 Male + Lee et al. 2005, Zegeer et al. 2006, Schneider et al. 2010, Jang et al. 2013, Chimba et al. 2014, Fitzpatrick et al. 2014, Palamara and Broughton 2013 Older adults + Lee et al. 2005, Zegeer et al. 2006, Schneider et al. 2010, Jang et al. 2013, Chimba et al. 2014, Fitzpatrick et al. 2014, Palamara and Broughton 2013 Children + Loukaitou et al. 2007, Jang et al. 2013, Chimba et al. 2014 Latino, Black and Hispanic + Lee et al. 2005 Alcohol and drug use + Martin 2006 Pedestrians with disabilities + Fitzpatrick et al. 2014 Weather and lighting Lighting − Lee et al. 2005, Jang et al. 2013, Fitzpatrick et al. 2014 Rainy weather + Jang et al. 2013, Fitzpatrick 2014 Night (time of day) + Performance Measures for Evaluating Operations and Mobility Pedestrian Speed Pedestrian speed is a critical issue in the design and operation of pedestrian facilities (Ishaque and Noland 2008). Walking speeds are influenced by environmental, traffic, and pedestrian characteristics, and differences in pedestrian speeds have been observed by age (Knoblauch et al. 1996, Gates et al. 2006, Fitzpatrick et al. 2006), gender (Fruin 1971, Polus et al. 1983, Tarawneh 2001, Montufar et al. 2007, Finnis and Walton, 2008), and group size (Tarawneh 2001, Carey 2005). Of these factors, age is the most significant. Pedestrian crossing speeds are important because they are used to set the pedestrian clearance time at signalized intersections and therefore impact intersection efficiency. Initially, the MUTCD recommended a walking speed of 4.0 ft/s for traffic signal timing (MUTCD 2000) based on LaPlante’s unpublished research from the 1950s, which showed an average walking speed of 4 ft/s for all crossing pedestrians irrespective of age and gender (LaPlante 2004). The data also showed a breakpoint in the speed frequency

177 distribution curves at the 15th percentile, whose value was 3.5 ft/s irrespective of age and 3.0 ft/s if elderly pedestrians were considered (LaPlante 2004). Knoblauch et al. (1996) found that elderly pedestrians are significantly slower than younger pedestrians (15th percentile speed for younger pedestrians (4.09 ft/s), 15th percentile speed for older pedestrians (3.19 ft/s). Older pedestrians were defined as those that appeared to be 65 years of age or older. Pedestrians who appeared to be under the age of 65 were classified as younger pedestrians. A TCRP–NCHRP study found 15th percentile walking speeds of 3.77 ft/s and 3.03 ft/s for younger and older pedestrians, respectively (Fitzpatrick et al. 2006). Subsequently, the MUTCD’s guidance was changed to a walking speed of 3.5 ft/sec for determining the pedestrian clearance time (MUTCD 2009). Table A4 lists the key studies and their recommendations with respect to walking speeds based on age. Table A4. Summary of Walking Speeds (ft/s) Based on Age. Study Overall Younger Pedestrians Older Pedestrians LaPlante 1950 3.5 3.0 Traffic Engineering Handbook 1982 3.7 3.3 Knoblauch et al. 1996 3.97 3.08 Coffin and Morrall 1996 4.0 (intersections); 3.3 (midblock crosswalks or at intersections near senior housing facilities) Guerrier and Jolibois 1998 3.09 3.31 2.20 Milazzo et al. 1999 4.0 (0% – 20% elderly pedestrians) 3.3 (greater than 20% elderly pedestrians) Traffic Control Devices Handbook 2001 4.0 3.5 Staplin et al. 2001 2.8 La Plante and Kaeser 2004 3.82 (problem intersections) Dahlstedt 1978 2.2 Bennett et al. 2001 4.0 Akçelik & Associates 2001 4.0 Fitzpatrick et al. 2006 3.77 3.03 Gates et al. 2006 3.8 3.6 (proportion of older adults exceed 20% of the total population) Sharifi et al. (2015) present findings of a laboratory study that measured the distribution of speeds of pedestrians with different kinds of disabilities and mobility devices (visual, nonmotorized wheelchair or rolling walker, motorized wheelchair, cane) in different walking environments (passageway, oblique angle, right angle, bottleneck, queuing area, stair) and levels of congestion (HCM LOS A/B, C/D, and E/F). Pecchini and Giuliani (2015) found that disabled pedestrians had slower crossing speeds compared to unimpaired pedestrians, and found that disability status, vehicle traffic characteristics (average gap), and infrastructure context are key factors that influence interactions between drivers and pedestrians. The impact on pedestrian speeds of activities performed during walking has also been studied. While Morall et al. (1991) found that people walking with luggage walked slower than pedestrians without luggage, Young (1998) found no significant differences between walking speeds of pedestrians with and without luggage. More recently, pedestrian distraction emerged as a factor that can influence walking

178 speeds. Alsaleh et al. (2018) evaluated the impact of pedestrian cell phone use on walking behavior and found that mean walking speeds between distracted and non-distracted pedestrians (5.44 ft/s vs. 4.88 ft/s) were significantly different. Russo et al. (2018) conducted an observational study exploring factors associated with distracted walking, pedestrian violations, and walking speed at signalized intersections, and found a mean walking speed of 4.8 ft/s across all study sites. Relationships between pedestrian speed and delay have also been studied. As delay at crossings increases, the pedestrian crossing speed also increases (Crompton 1979). Pedestrian delay is also associated with level of traffic flow (Hine and Russell 1996). The choice of gap acceptance also impacts pedestrian speeds. Researchers have found that pedestrians who accept shorter gaps have higher walking speeds (Moore 1953; DiPietro and King 1970, Cohen et al. 1955). While some studies have found that males accept smaller gaps than females (Cohen et al. 1955, DiPietro and King 1970, Wilson and Grayson 1980), other studies have found that children and young adults accept smaller gaps than older adults (Das et al. 2005). Pedestrian crossing speed is also dependent on the time in the signal cycle during which a pedestrian arrives at the intersection or crossing (Ishaque and Noland, 2008). Virkler found that pedestrians arriving during the pedestrian clearance phase may choose to cross the street during the clearance phase at an increased speed, rather than wait for the next cycle (Virkler 1998). Some studies have found higher pedestrian speeds during pedestrian clearance and red phases (Lam et al. 1995, Gates et al. 2006). Pedestrian speeds on sidewalks are strongly influenced by the presence of other pedestrians (Ishaque and Noland 2008). On sidewalks, Willis et al. (2004) found that variation in speeds is higher for males and children than females and elderly adults. Pedestrian Speed–Flow–Density Speed–density relationships have been derived for pedestrian flows and are generally described by linear relationships (HCM 1985, Fruin 1971, Pushkarev and Zupan 1975, May 1990, Tanaboriboon and Guyano 1989, Tanaboriboon et al. 1986). Polus et al. (1983) modeled the relationship using a multi-regime linear model. The basic relationship between pedestrian flow, speed, and density is given by 𝐹𝑙𝑜𝑤 (𝑞) = 𝑠𝑝𝑒𝑒𝑑(𝑣) × 𝑑𝑒𝑛𝑠𝑖𝑡𝑦(𝑘) where 𝑞 is the pedestrian flow rate, number of pedestrians walking across a unit width of path in unit time (p/min/ft), 𝑣 is the average pedestrian speed (ft/min), and 𝑘 is the density, the number of persons in a unit area (p/ft2). Ishaque and Noland (2008) suggest that while most of the fitted curves describing the relationship are linear, the variation with some non-linear fits arises due to different trip purposes. Navin and Wheeler (1969) measured the effects of bi-directional pedestrian flows on sidewalk capacity of the sidewalk using time-lapse color photos. Virkler and Elayadath (1994) estimated seven models to describe the speed–flow– density relationships. Differences in speeds due to trip purposes were seen in the curves produced by Navin and Wheeler (1969), where young students were hurrying to class, and in those produced by Virkler and Elayadath (1994), where pedestrians with non-urgent trip purposes were moving slowly. Some studies estimated a linear relationship between speed and density for a single regime (Older 1968, Navin and Wheeler 1969, Fruin 1971, Tanaboriboon et al. 1986, Tanaboriboon and Guyano 1989, Gerilla 1995; Lam et al. 1995, Sarkar and Janardhan 2001, Al-Azawwi and Raeside 2007, Kotkar et al. 2010, Nazir et al. 2012, Rastogi et al. 2013). Others proposed relationships for two regimes (i.e., two density ranges) (Virkler and Elayadath 1994, Lee 2005) and three regimes (Polus et al. 1983, Jia et al. 2009). Some researchers have also found capacity losses resulting from bi-directional flows (Navin and Wheeler 1969, Fruin 1971, Al-Masaeid et al. 1993). A summary of the studies and the relationships is shown in Table A5.

179 Table A5. Summary of Speed–Density Studies. Study Location Flow Direction Regimes Speed–Density Equation(s) Hankin and Wright (1958) Experimental, UK Not defined Older (1968) Sidewalks, UK Single v = 78.6 – 20.4k Navin and Wheeler (1969) Sidewalks, USA Bi Single v = 127.8 – 47.4k Fruin (1971) Transport terminal and stairways, USA Uni & bi Single v = 85.8 − 21k Tanaboriboon et al. (1986) Sidewalks, Singapore Bi Single v = 75.6 − 15.6k Tanaboriboon and Guyano (1989) Sidewalks, Singapore Bi Single v = 72.85 − 13.13 k Weidmann (1993) Stairways and level surface, Switzerland Uni & bi Single v = 80.40[1 – e(−1.913{(1/k)−(1/5.4)})] (level) v = 36.6[1 – e(−3.722{(1/k)−(1/5.4)})] (up) v = 41.64[1 – e(−3.802{(1/k)−(1/5.4)})] (down) Virkler and Elayadath (1994) – – Two v = 60.6 e−(0.24k) (k < 1.07) v = 36.6 ln (4.32/k) (k >1.07) Polus et al. (1983) Sidewalks, Israel Uni Three v = 1.272 – 0.122k, k < 0.6 v = 1.623-0.732k, 0.61 < k < 0.75 v= 1.326-0.273k, k > 0.75 Al-Masseid et al. (1993) Carriageways, Jordan Bi Single v = (19.2 + 123v – 93v2)/k Gerilla (1995) Sidewalks and crosswalks, Philippines Bi Single v = 83.23 − 23.1k Lam et al. (1995) Crosswalks, walkways, escalators and stairways, Hong Kong Uni Single v = 1.29 − 0.36k (indoor walkway) v = exp(0.38 − 0.57k) (outdoor walkway) v = 1.42 e(−0.347k^2) (signalized crossing) v = 1.67 e(−0.5k) (signalized crossing) Sarkar and Janardhan (2001) Transfer terminal, India Bi Single v = 87.6 − 21k. Lee (2005) Public transport facilities, Netherlands Uni Two v = 46.2 if k ≥ 0.6 v = 46.2 (1.125 − 0.208k) if 0.6, k ≤ 5.4. (down stairways) v = 40.8 if k ≥ 0.6, v = 40.8 (1.125 − 0.208k) if 0.6 , k ≤ 5.4. (up stairways) v = 52.8 if k ≥ 0.6 v = 52.8(1.06 − 0.1k) if 0.6, k ≤ 4.8 (down escalator) v = 49.2 if k ≥ 0.6, v = 49.2(1.07 − 0.12k) if 0.6 , k , 3.8 (up escalator) Al-Azzawi and Raeside (2007) Shopping and urban business areas, U.K. Bi Single ln v = 0.674 ln q − 0.448 ln k Jia et al. (2009) Transport terminal, China Uni Single v = −35.34k + 101.22 (corridors) v = −7.92 ln k + 41.4 (up stairs) v = −10.5 k + 56.46 (down stairs) Chattaraj et al. (2009) Experimental, India Uni NA

180 Study Location Flow Direction Regimes Speed–Density Equation(s) Kotkar et al. (2010) Carriageway and sidewalk, India Bi Single v = 82.21 − 24.47k (up, location 1) v = 83.25 − 24.44k (down, location 1) v = 88.93 − 24.98k (up, location 2) v = 86.36 − 19.48k (down, location 2) v = 82.52 − 31.85k (up, location 3) v = 81.07 − 31.75k (down, location 3) v = 83.79 − 20.05k (up, location 4) Nazir et al. (2012) Sidewalks, Bangladesh Uni Single v = 64.14 − 9.97k (location 1) v = 61.19 − 8.16k (location 2) v = 62.89 − 9.2k (location 3) Rastogi et al. (2013) Sidewalk precincts and carriageway, India Bi Single v = 94.56e−k/3.03 (sidewalks) v = 89.52e−k/2.857 (wide sidewalks) v = 80.4e−k/2.564 (precincts) Source: Adapted from Gupta and Pandir (2015) and Ishaque and Noland (2008). Note: Relationships are presented in metric units, v: m/min, k: ped/m2 The majority of studies in the literature have been conducted in developed countries and there is lack of research on pedestrian flow characteristics in developing countries (Seyfried et al. 2009). As both a lack of exclusive facilities and crowds affect pedestrian behavior, results of studies conducted in developed countries may not be directly transferrable to developing countries. Fruin (1971) and Helbing et al. (2007) suggest that the variations in results may be due to cultural differences. More research is still needed on effects of grade on pedestrian flow and on bottlenecks at pedestrian facilities, although an old study performed in Pittsburgh and referenced in the HCM investigated the effect of grade on pedestrian speed (Municipal Planning Association 1923). Gupta and Pundir (2005) also suggest the need for an integrated approach for studying pedestrian flow characteristics considering all factors that impact flows. Pedestrian Satisfaction Planners have long studied the relationship between walking behavior and the built environment. The design quality of streetscapes also influences walkability and reflects pedestrian perceptions regarding the environment (Ewing et al. 2006). A list of perceptual qualities described in the literature include adaptability, ambiguity, centrality, clarity, compatibility, comfort, complementarity, continuity, contrast, deflection, depth, distinctiveness, diversity, dominance, expectancy, focality, formality, identifiability, intelligibility, interest, intimacy, intricacy, meaning, mystery, naturalness, novelty, openness, ornateness, prospect, refuge, regularity, rhythm, richness, sensuousness, singularity, spaciousness, territoriality, texture, unity, upkeep, variety, visibility, and vividness (Ewing et al. 2006). A complete review can be found elsewhere (Ewing 2000, Ewing et al. 2005). Ewing et al. focused on eight urban design qualities— imageability, legibility, visual enclosure, human scale, transparency, linkage, complexity, and coherence— to determine their impacts on walkability (Ewing et al. 2006). The methodology consisted of a combination of rating of urban design qualities of streetscapes by the expert panel and measurement of physical features of streetscapes through a content analysis of video clips. The findings revealed that human scale, transparency, enclosure, and imageability had the largest impact on walkability (Ewing et al. 2006). A number of studies have explored pedestrian satisfaction with the walking environment, which consists of sidewalks and crosswalks. These studies typically use either web-based or intercept surveys to determine pedestrian level of satisfaction with various factors that can influence their perception of the walking experience. The levels of satisfaction are measured using Likert or other scales; these are sometimes converted to quantitative LOS measures and used to rank different facilities. Factors impacting walkability can be broadly divided into the following categories.

181  Physical infrastructure – Sidewalk presence – Sidewalk width – Sidewalk continuity – Slope – Bus shelters – Parking – Crosswalk presence – Pedestrian signal – Median island  Road Safety – Vehicular volume – Traffic noise – Traffic fumes – Pedestrian flow rate – Waiting time – Crossing distance  Aesthetics – Sidewalk cleanliness – Sidewalk surface quality and evenness – Obstructions (trash cans, light poles, sign posts, etc.) – Trees and other greenery  Access and Facilities – Disabled pedestrian access – Land use mix  Security – Street lighting – Cameras – Police patrols Many studies have found that pedestrians highly value attributes such as road width, vehicle speed and volumes, connectivity, and lighting conditions that directly impact safety, compared to attributes that are related to comfort or convenience (Araujo et al. 2008, Cleland et al. 2008, Hung et al. 2010, Martinez and Barros 2014, Said et al. 2016, Rajendran et al. 2018). Differences in satisfaction based on gender (Arshad et al. 2016, Parida et al. 2011) and trip purpose (Hong and Park 2017, Kothuri et al. 2014) were observed. Satisfaction at crosswalks was dependent on lower delays and shorter crossing times (Jensen, 2013). Younger adults, people making short-duration trips, and those using public transportation were also less satisfied with intersection delays (Kothuri et al. 2014). Table A6 summarizes the key studies and their findings.

182 Table A6. Summary of Studies on Pedestrian Satisfaction. Study Facility Factors Method Location Key Findings Tokunaga et al. 2004 Sidewalks, Crosswalks Walking location, icy/snowy, evenness, sanding Pedestrian satisfaction level with different sidewalk conditions was evaluated through a self-administered survey based on the conjoint approach. Sapporo, Japan Icy or snowy sidewalks are the biggest barrier for walking. Icy, uneven surfaces between crosswalk and sidewalks without antislipping materials were difficult to walk on. Hung et al. 2010 Walkability Street lighting, sidewalk width, sidewalk grade, sidewalk cleanliness, obstacles, crossing point frequency, access, vehicular traffic, vehicular speed, weatherproof facilities Pedestrian interview survey, field walkability survey Hong Kong Walkways in commercial areas have better infrastructure and higher LOS. Improvements in street lighting; clean, weatherproof, and wider sidewalks; reducing traffic volumes and speed; removal of obstacles on sidewalks; and higher number of crosswalks were desirable factors to improve pedestrian satisfaction. Parida et al. 2011 Sidewalks Sidewalk width, sidewalk surface, obstruction, security, and comfort Questionnaire-based survey for qualitative assessment of sidewalk facilities; a form with a rating scale was used for qualitative evaluation of pedestrian facilities. Total weighted score was used to calculate LOS Delhi, India Physical design of sidewalks is not sensitive to gender difference; however, differences are pronounced for security. Araujo et al. 2008 Crosswalks Satisfaction with performance attributes such as comfort, safety, and system continuity was measured. Performance measures such as waiting time, space available while waiting to cross, number of pedestrians, one-way or two-way, state of road surface, road width, vehicular speed, visibility, lighting conditions, guardrails, absence of obstacles in the vicinity, sidewalk condition, lowered curb, pedestrian signals, and central island were considered. Performance measures (PMs) were selected and psychometric procedures were applied in a field study with pedestrians to obtain weights for each PM considered. The final list of performance measures was derived and user satisfaction for each PM was measured by correlating against LOS. The resulting qualitative LOS for each facility was then determined. São Paulo, Brazil Pedestrians give highest priority to safety attributes such as road width, vehicular speed, visibility, lighting conditions, and guardrail presence. System continuity factors such as obstacle presence in the vicinity of a crossing, sidewalk condition, lowered curb, pedestrian signals and median island existence were next. In last place was comfort and its attributes. Perceived LOS did not match the quantitative LOS estimated using average pedestrian delay. Cleland et al. 2008 Walkability Aesthetics, physical infrastructure, road safety, access and facilities, social environment. Survey of mothers of elementary school children Melbourne, Australia Connectivity, lighting, safety (traffic speed and presence of traffic calming devices), greater satisfaction with local facilities and close-knit neighborhood were factors that were positively associated with walking for transport (commute). Supportive social environment, satisfaction with the quality of local facilities, and connectivity were associated with high rates of leisure walking.

183 Study Facility Factors Method Location Key Findings Sahani et al. 2017 Crosswalks Number of lanes crossed, 85th percentile vehicle speed, delay, number of left-turning vehicles, number of permissible right-turning vehicles, number of through vehicles, number of left/right/through nonmotorized vehicles Qualitative data collected using a perception survey, SOM in ANN clustering approach was used to define ranges of PLOS scores for six categories Eight mid- size cities in India Total pedestrian delay increases linearly with increase in waiting time delay. PLOS models show that at intersections with pedestrian delay < 20 sec provide the best QOS and the QOS decreases with increasing pedestrian delay. Increase in turning vehicle volumes leads to lower satisfaction. Survey suggested that pedestrians want exclusive pedestrian phases to cross. Jung et al. 2016 Street Design street, year, sidewalk width, lane width, number of lanes, bus lanes, presence of separated sidewalk, bus stops within 50 m, subway stations within 50 m, presence of trees Pedestrian satisfaction survey using a 5-point Likert scale Seoul, South Korea Pedestrians walking on facilities with street-improvement projects were more satisfied compared to those on typical streets. Presence of trees was a significant factor that improved satisfaction, whereas other factors such as bus stops, sidewalk width, and presence of street lighting were not significant. Pedestrian volumes were higher on streets with bus- dedicated lanes and subway stations and in places with higher employment density. Wang et al. 2012 Sidewalks Sidewalk tree, surface condition, type of land use, shrubs, signage and sidewalk width. Perception survey using slides Iksan, South Korea Overall satisfaction is more directly affected by perception factors than physical components. Two emotional perception factors—naturalness and openness—were used in this study. Harmoniousness is positively correlated with existence of sidewalk trees, existence and height of shrubs, and amount of signage. Openness is also positively correlated with all physical components although it is strongly correlated with sidewalk width. Harmoniousness has a stronger association with overall satisfaction than openness. Arshad et al. 2016 Sidewalks Overall travel experience, sidewalk condition, crossing, sidewalk amenities, personal safety, adjacent traffic/driver behavior, aesthetics and amenities, accessibility Questionnaire survey Kuala Lumpur, Malaysia Differences in satisfaction between male and female pedestrians were observed. Both genders were concerned about safety while walking.

184 Study Facility Factors Method Location Key Findings Hong and Park 2017 Sidewalks Trip attributes (trip frequency, time of day), trip purpose, sidewalk environment (crosswalk, slope, fence, bus stop, subway stop), obstacles (tree, light pole, street stall, traffic control device, trash can, subway entrance, illegal parking), land use, pedestrian characteristics (age, employment, gender) Intercept pedestrian survey with individual interviews of people. Mixed process ordered probit and two-stage residual inclusion ordered probit models were built. Seoul, South Korea Pedestrian satisfaction varies by trip purpose; shopping, business, and social activity trips had better levels of satisfaction than commute and exercise trips. Existence of light poles, traffic signal poles, street stalls, shop display stands, bus stops, and subway entrances improved satisfaction. Higher numbers of crosswalks and sidewalks were associated with lower satisfaction rates. Kothuri et al. 2014 Crosswalks Demographics (age, gender), trip characteristics (groups, public transit availability, trip purpose, trip length, trip frequency), perceptions (delay, safety), attitudes (compliance, safety) Intercept survey followed by binary logit models Portland, OR Safety was of bigger concern than compliance while crossing. Trip purpose is associated with safety, people on work trips and making longer-duration trips were less concerned with safety while crossing. People crossing in groups are also less concerned about safety. Younger adults, people on short-duration trips, and those made using public transportation are less satisfied with delay. Intersections on pedestrian recall had higher satisfaction with delays compared to actuated intersections. Kim et al. 2014 Meso-scale (floor area ratio, intersection density, slope, subway, bus), micro-scale (bus lane, crossing, sidewalk width, fence, signal, lamp, ramp, tree, trash can), socio- demographic (groups, familiarity with environment, gender, age) Survey of pedestrians followed by multilevel model estimation Seoul, South Korea Utilitarian pedestrians are more likely to be satisfied when higher density and public transport are available. Negative association was found between pedestrian satisfaction and intersection density. Grade was also negatively correlated with satisfaction. Meso-scale variables such as dedicated bus lanes, availability of crossings, wider streets, and trees are positively associated with satisfaction. For both utilitarian and recreational models, meso- scale variables were more important than micro-scale variables. Ghani et al. 2017 Traffic function - Safety (separation, traffic flow), security (conflict point, protection from crime), accessibility (mixed building use, inclusive design), facility (street furniture), mobility (continuous sidewalk, pedestrian flow). Tourism function - Attractiveness, Enjoyment, Convenience, Comfort Questionnaire survey Melaka, Malaysia Pedestrian satisfaction index is developed which is composed of traffic and tourism functions in a street.

185 Study Facility Factors Method Location Key Findings Zhao et al. 2015 Sidewalks Pedestrian flow rate, bicycle volume, electric bike volume, vehicle volume in the inside lane, sidewalk width, frequency of barriers on sidewalks, sidewalk environment Video observation and intercept survey, followed by modeling using fuzzy neural networks Beijing, China When pedestrian flow rates were low, road facility conditions and environmental factors affected ped LOS, however the impact of these factors weakened as the pedestrian flow rate increased to high-density conditions. As pedestrian flow rates increased, the LOS was improved by providing separate facilities for pedestrians and vehicles. Rajendran et al. 2018 Sidewalks Mobility and infrastructure (good sidewalk surface, wider sidewalks, presence of bus shelters, continuous sidewalks, absence of encroachment, pedestrian amenities); comfort and convenience (disability access, absence of obstructions, cleaner sidewalks); security (police patrols, cameras, street lighting); safety (traffic speed, traffic volume) Survey followed by structural equation modeling Thiruva- nantha- puram, India Security construct has the largest effect on PLOS, followed by factors associated with comfort and convenience. Factors associated with mobility and infrastructure had the lowest level of pedestrian satisfaction. Said et al. 2016 Walkability Sidewalk width, sidewalk surface quality, diversity of activities, trees, proportion of shadowed sidewalk, buildings with entrance door access to sidewalk, parking access across sidewalk, bollards, parking meters, sign posts, cars and motorcycles parked on sidewalk, sidewalk infringement, electricity poles, trash bins Web-based survey of university students, followed by structural equation modeling Beirut, Lebanon Level of satisfaction with walking environment was most influenced by neighborhood attributes (ease of pedestrian crossing, sidewalk blockage, sidewalk cleanliness, vehicular traffic levels, and motorcycles riding against traffic on one-way streets), followed by diversity of activities. Sidewalk width and quality were not significant. For commute walking trips, correlations were found between level of satisfaction with the walking environment and level of satisfaction with the commute. Jensen 2013 Crosswalks Signalized intersections (crosswalk and pedestrian signal, median island, waiting time, land use, crossing distance, motor vehicles on crossing and parallel to crossing); Roundabouts (crosswalk and land use, circulating lane and splitter island, crossing distance, circulating motor vehicles and motor vehicles on crossed arm); Non- signalized crossing (pedestrian facility, average speed of motor vehicles on crossed road, crossing distance, motor vehicles on crossed road) Video clips were shown to a random sample of participants and crossings were rated on a six-point scale. Cumulative logit regression models of pedestrian and cyclist satisfaction were developed Denmark At signalized intersections, presence of marked crosswalks was the most significant factor for pedestrian satisfaction. Lower delays and shorter crossing times also contributed to higher levels of satisfaction. Higher traffic volumes were also associated with higher levels of satisfaction. At non-signalized intersections, marked crosswalks positively influence pedestrian satisfaction; however, their effect is smaller when compared to a signalized intersection. Type of facility that a pedestrian walks on before reaching the crossing has a higher impact on pedestrian satisfaction. Traffic volumes have a negative impact on pedestrian satisfaction.

186 Study Facility Factors Method Location Key Findings Martinez and Barros 2014 Walkability Walking impedances (sidewalk quality, presence of barriers, street lighting, parking); Topoceptive awareness (land use diversity, presence of open spaces, presence of walls or windows, building height, presence of trees, block length, land use, road hierarchy); Safety and security (distance between pedestrian crossings, presence of public transport, street lighting, traffic intensity); Urban design (slope, sinuosity, land use, road hierarchy); Walking comfort (street width, sidewalk width, tree presence, traffic segregation); Traffic typology (traffic intensity, type of movement, traffic segregation) Web-based survey followed by structural equation models Portugal, Brazil Impedance (pedestrian infrastructure) is the most significant factor that contributes to pedestrian satisfaction, followed by factors related to walking comfort, safety and security. Safety and security are more significant in Latin America and the Caribbean, compared to Europe Pedestrian Safety Countermeasure Effects on Pedestrian Safety, Operations, and Quality of Service Commonly Used Pedestrian Safety Countermeasures To improve the safety of the roadway for pedestrians, various countermeasures have been proposed and implemented. These countermeasures are typically either targeted at roadways or intersections, and can also aim at reducing traffic or travel speed. A comprehensive list of countermeasures used to improve pedestrian safety can be found in PEDSAFE (Zegeer et al. 2013). Table A7 summarizes the most commonly used countermeasures. Effects of Countermeasures on Safety, Operations, and Quality of Service A number of countermeasures have been deployed by agencies to reduce pedestrian–vehicle conflicts and improve safety. Mead et al. (2013); Kittelson & Associates, Inc. et al. (2017); and Monsere et al. (2017) specifically list safety and operational impacts of various pedestrian-focused safety countermeasures by intersections and crossing treatments. Table A8 lists the factors affecting pedestrian safety at intersections, along with their impacts on safety, while Table A9 lists the countermeasures employed at crosswalks and their impacts. In general, most of the countermeasures decrease pedestrian–vehicle conflicts and increase driver yielding rates, although the magnitudes vary between studies.

187 Table A7. List of Commonly Used Pedestrian Countermeasures. Factors Description Roadway Segments Sidewalks Pedestrian lanes that provide people with dedicated right-of-way Signalized Intersections Red clearance interval All-red phase at the intersection signal between phase changes Exclusive phasing/pedestrian scramble An exclusive pedestrian phase such that pedestrians can cross at any direction, including diagonal Leading pedestrian interval Programmed into traffic signals to give the pedestrians the WALK signal 3–7 seconds before the motorists are allowed to proceed through the intersection Pedestrian push buttons Pedestrian push buttons to activate pedestrian phases Pedestrian countdown timers Countdown timers to inform pedestrians of duration left until green pedestrian phase Curb extensions/parking removal near the intersection An extension to the sidewalk into the parking lane at intersections and restriction of parking near the intersection Curb ramps Ramps to provide access between sidewalk and roadway Unsignalized Crosswalks Marked crosswalks Indicates optimal or preferred locations for pedestrians to cross and communicate to motorists that they should yield to pedestrians Advance yield/stop signs A tiled or stop sign placed 20–50 feet ahead of the crosswalk High-visibility crosswalks Crosswalk markings making use of continental style pavement markings Pedestrian hybrid beacons Includes an overhead beacon with three sections and signs labeled CROSSWALK STOP ON RED, along with a marked crosswalk on the major street and countdown pedestrian signal heads. Rectangular rapid-flashing beacons Uses a fast rapid-pulsing flash rate with a bright light intensity, placed on both sides of a crosswalk with a pedestrian crossing sign. The flashing patterns is activated with push-button or automatic detection of pedestrian. Full pedestrian signal A full red–yellow–green display for traffic. Signal may be activated with push-button, recall, or automatic pedestrian detection Median refuge Raised islands placed in the center of the street at intersections or midblock crossings Raised median Curbed sections that typically occupy the center of the roadway that can facilitate pedestrian crossing by providing a crossing area that is physically separated from the vehicular travel way. Lighting and illumination Appropriate quality and placement of lighting to improve nighttime visibility Pedestrian overpasses/underpasses Allows for uninterrupted flow of pedestrian movement separate from vehicle traffic Raised pedestrian crossings A speed table covering an entire intersection or midblock crossing

188 Table A8. Factors Affecting Pedestrian Safety at Intersections. Factors Study Key Findings Number of lanes and approach legs Nabavi et al. 2016, Thomas et al. 2017 Positively correlated with pedestrian crash risk Red clearance interval Retting et al. 2002 Implementation reduced pedestrian and bicycle crash risk by 37% Parking removal near intersection Shawky et al. 2014, Thomas et al. 2016, Obeid et al. 2017 Reduced pedestrian crashes by 30% Signage (NO TURN ON RED, STOP HERE ON RED, RIGHT ON RED ARROW AFTER STOP, TURNING VEHICLES YIELD TO PEDESTRIANS) City of Madison 1999, Huang et al. 2000, Kamyab et al. 2003, Strong et al. 2006, Banerjee et al. 2007, Ellis et al. 2007, Hua et al. 2009, Pécheux et al. 2009, Bennett et al. 2014, Gedafa et al. 2014 Statistically significant increase in driver yielding behavior and number of pedestrians diverting to treated crosswalks, lower vehicle speeds, increased right- turn driver compliance, highest compliance observed for NO TURN ON RED sign Exclusive phasing (pedestrian scramble) Zegeer et al. 1985, Bechtel et al. 2004, Kattan et al. 2009, Chen et al. 2013, Ivan et al. 2017 Mixed impacts, statistically significant reduction in pedestrian–vehicle conflicts, decrease in pedestrian compliance Leading pedestrian interval King 1999, Van Houten et al. 2000, Hua et al. 2009, Pécheux et al. 2009, Fayish and Gross 2010 Statistically significant reduction in vehicle–pedestrian conflicts, reduction in conflict severity, increase in number of pedestrians who used the push button and crossed during the walk phase Addition of pedestrian push buttons Hughes et al. 2000, Van Houten et al. 2006, Bradbury et al. 2012 Increase in pedestrian compliance and reduction in pedestrian crashes Automated pedestrian detection Hughes et al. 2000, Pécheux et al. 2009, Nambisan et al. 2009, Lovejoy et al. 2012 Statistically significant reductions in pedestrian–vehicle conflicts, number of people entering the crosswalk during the DON’T WALK phase, and late crossings Pedestrian countdown timers Eccles et al. 2004, Markowitz et al. 2006, Schrock et al. 2008, Reddy et al. 2008, Camden et al. 2012, Vasudevan et al. 2011, Levasseur et al. 2011, Sharma et al. 2012, Van Houten et al. 2012, Van Houten 2014 Statistically significant reduction in pedestrian–vehicle conflicts, safer speed decisions when approaching the intersection, statistically significant increase in successful crossings, improved pedestrian crossing behavior, faster walking speeds and increase in pedestrian compliance.

189 Table A9. Key Factors Impacting Pedestrian Safety at Crosswalks. Factors Studies Key Findings Marked crosswalks Zegeer et al. 2002, Mitman et al. 2008 At uncontrolled locations on two-lane roads or on multilane roads with low ADT, no statistical difference in pedestrian crash rate with marked crosswalks. On multilane facilities with higher ADT and marked crosswalk alone, without other enhancements, led to statistically higher pedestrian crash rate. Statistically significant higher yielding rates at marked crosswalks, compared to unmarked crosswalks. Curb extensions King 1999, Van Hengel 2013, Bella and Silvestri 2015 Reduction in overall crash severity rate, reduction in driver speed at they approached the crossing, increase in far-lane driver yielding rates Advance yield markings Van Houten 1988, 1992, 2001, 2002; Malenfant et al. 1989; Nambisan et al. 2007; Van Hengel 2013; Decrease in vehicle–pedestrian conflicts, increase in driver yielding behavior at T-intersections, most effective when combined with other treatments such as refuge islands High-visibility crosswalks Nitzburg et al., 2001, Feldman et al. 2010, Fitzpatrick et al. 2011, Chen et al. 2013, Mead et al. 2014 Increased yielding rates, decrease in collision rates, increase in pedestrians looking for vehicles before crossing, and increase in yielding distance. Detection distances for continental and bar pairs were similar and statistically longer than for transverse markings during day and night. Pedestrian hybrid beacons Fitzpatrick et al. 2006, Fitzpatrick et al. 2010, Fitzpatrick et al. 2014, Pulugurtha et al. 2013, Gates et al. 2016, FHWA 2016 Increased yielding rates on facilities with multiple lanes, statistically significant reduction in total crashes, and higher reductions in pedestrian– vehicle crashes, reduction in proportion of trapped pedestrians Rectangular rapid- flashing beacons Van Houten et al. 2008, Pécheux et al. 2009, Hua et al. 2009, Hunter et al. 2009, Shurbutt et al. 2010, Ross et al. 2011, Domarad et al. 2013, Fitzpatrick et al. 2014, Foster et al. 2014, Mead et al. 2014, Mishra 2015, Gates et al. 2016, FHWA 2016, Dougald et al. 2016, Fartash et al. 2016, Fitzpatrick et al. 2017, Zegeer et al. 2017, Monsere et al. 2017 Statistically significant increase in driver yielding behavior, reductions in pedestrian–vehicle crash risk and trapped pedestrians

190 Factors Studies Key Findings Median refuge Bowman et al. 1994, Bacquie et al. 2001, Huang et al. 2001, Kamyab et al. 2003, King et al. 2003, Zegeer et al. 2005, Pécheux et al. 2009, Pulugurtha et al. 2012, Zegeer et al. 2017, Fitzpatrick et al. 2017 Most studies showed positive safety benefits such as lower pedestrian crash rates, significant increase in pedestrians using the crosswalk, significant increase in proportion of drivers yielding to pedestrians, significant reduction in mean speeds and increase in speed limit compliance. One study found no statistically significant improvements in driver yielding, number of trapped pedestrians or pedestrian– vehicle conflicts at two sites (Pécheux et al. 2009) Raised median Huang et al. 2001, Schneider et al. 2010, Thomas et al. 2016 Statistically significant reduction in speeds, statistically significant increase in driver yielding behavior, increase in number of pedestrians using the crosswalk, reduced pedestrian crash risk Lighting Nambisan et al. 2009, Yuan et al. 2017 Statistically significant increase in crosswalk utilization and increase in driver yielding rates, statistically significant decrease in the proportion of pedestrians trapped in the roadway Pedestrian-activated flashing yellow beacons Nitzburg et al. 2001, Huang et al. 2000, Van Houten et al. 1998, Pécheux et al. 2009, Hua et al. 2009 Increased driver yielding rates, often used in conjunction with other treatments such as illuminated signs, high-visibility crosswalks and advanced yield markings. In-pavement lighting Godfrey et al. 1999, Huang et al. 2000, Hakkert et al. 2002, Van Derlofske et al. 2003, Karkee et al. 2006, Mead et al. 2014 Mixed results, some studies reported increase in driver yielding rates, reductions in vehicle speeds and collision rates, while others reported the opposite effect. Pedestrian overpasses and underpasses Campbell et al. 2004 Substantial reductions in pedestrian crossing collisions, but increase in other crashes Crash Modification Factors (CMFs) have been estimated to quantify the impacts of selected countermeasures on pedestrian safety. A crash modification factor is a numerical estimate of the expected reduction (or increase) in the number of crashes that may result when a countermeasure treatment is implemented. The CMF Clearinghouse (www.cmfclearinghouse.org) provides CMFs for a variety of countermeasures along with a star rating that indicates the robustness of the study. NCHRP Project 17-73 (2018) provides a summary of pedestrian countermeasures and their CMFs; a condensed table is reproduced as Table A10. NCHRP Project 07-25 (Kittelson & Associates, Inc. et al. 2017) conducted a literature review on alternative intersection and interchange (AII) forms (e.g., displaced left-turn, restricted crossing U-turn, diverging diamond) that included an element addressing pedestrian safety. Although research has developed CMFs for some AII forms showing reductions in overall crashes, no pedestrian-specific CMFs for AIIs were identified. The review concluded that “for many AII designs, shorter, multiple-phase crossings may be useful for limiting pedestrian exposure, promoting operational efficiency, and promoting wayfinding, at the cost of potentially increased pedestrian delay.”

191 Table A10. Summary of Countermeasure CMFs. Countermeasure Pedestrian CMF Motor Vehicle CMF High-visibility crosswalk 0.52 (Chen et al. 2013) 0.63 (Feldman et al. 2010) 0.81 (angle, head-on, left turn, rear end, rear to rear, right turn, sideswipe) (Chen et al. 2012) Raised crosswalk or speed table 0.55 (Elvik and Vaa 2004) 0.70 (serious, minor and possible injuries) (Elvik and Vaa 2004) Median crossing island 0.68 (Zegeer et al. 2017 a,b) 0.54-0.69 (Alluri et al. 2012, 2013, Zegeer et al. 2002) 0.71-0.74 (Zegeer 2017a) Road diet Not available; injury reductions expected due to lower speeds, fewer lanes (Thomas et al. 2016) 0.71 average/suburban roads 0.53 (suburban area) 0.81 (urban area) (Harkey et al. 2008) Curb extension with parking restriction 0.7 for parking removal (Gan et al. 2005) CMF not available for curb extension Not available Improve lighting 0.58 nighttime (Elvik and Vaa 2004) 0.77 (total injury crashes, CMFs for other crashes available in the clearinghouse) (Harkey et al. 2008) RRFBs 0.53–0.64 0.93 In-roadway YIELD FOR PEDESTRIAN sign (R1-6) Not available Not available Advance stop/yield marking and R1-5/R1-5a signs 0.75 0.64–0.86 (Zegeer et al. 2017a,b) 0.89 (total crashes) 0.80 (rear end and sideswipe) (Zegeer et al. 2017) Pedestrian hybrid beacon 0.31 (Fitzpatrick and Park 2010) 0.45 (Zegeer et al. 2017 a,b) 0.43 (PHB+Advance Stop/Yield) (Zegeer et al. 2017 a,b) 0.71 total crashes 0.85 fatal, serious injury (Zegeer et al. 2017 a,b) LPIs 0.41–0.95 (FHWA 2004, Fayish and Gross 2010, Brunson 2017) Not available Longer pedestrian phase 0.50 (Chen et al. 2014) 0.98 (all multi-vehicle crashes) (Chen et al. 2013) Protected crossing phase 0.61 (urban intersections) 0.49 (Chen et al. 2014) 0.01 left-turn crashes for restricted left (Harkey et al. 2008) Source: Adapted from Thomas et al. (2018) and Monsere et al. (2017). NCHRP Project 03-120 (Kittelson & Associates, Inc. et al. 2018) explored the relationship between access management techniques and their impacts on multimodal users along corridors. Although a number of techniques were identified, only few have documented, quantifiable impacts on pedestrian safety. These include:  Installing a non-traversable median on an undivided highway and replacing a two-way left-turn lane with a non-traversable median  Installing a right-turn deceleration lane  Installing channelizing islands to move the ingress merge point laterally away from highway Other techniques, such as establishing traffic signal spacing criteria, establishing spacing for unsignalized access, consolidating driveways, restricting parking on the roadway next to the driveway to increase

192 driveway turning speeds and improve sight distance, installing roundabouts, and installing driveways with the appropriate return radii, throat width, and throat length for the type of traffic to be served showed trends which were qualitative in nature. A number of other techniques are also listed in the project’s guidebook with possible impacts on pedestrian safety, however no information was found in the literature regarding definite impacts (Kittelson & Associates, Inc. et al. 2018). Dutata et al. (2014) investigated factors that led to pedestrian noncompliance with traffic signals. They found that noncompliance increased as pedestrian delay increased and at intersections with protected left- turn phases. Noncompliance decreased with longer crosswalks and higher conflicting traffic volumes. They note that narrowing the street cross-section to provide a shorter crosswalk may offset some of the safety benefit obtained from a shorter cycle length made possible by the shorter crossing distance. Anciaes and Jones (2018) investigated pedestrian preferences for using pedestrian over- or under- crossings instead of a signalized crossing. Depending on the model form used, pedestrians would have to walk 17–21 minutes farther to use a signalized crossing before they would choose a grade-separated crossing. Women would walk farther than men to avoid a grade-separated crossing, persons aged 50–65 would walk further than those aged less than 50 to avoid a grade-separated crossing, and persons aged greater than 65 would walk further than those aged 50–65 to avoid a grade-separated crossing. The authors also found no significant difference in stated pedestrian preference for “staggered” crossings at signalized intersections (i.e., incorporating a median refuge) and a “straight” signalized crossing (i.e., with no refuge). Current HCM Methods for Evaluating On-Street Pedestrian Operations and Quality of Service This section describes the current methods used in the HCM 6th Edition for analyzing pedestrian operations and QOS, the research basis for the methods, the sensitivity of the LOS result to the required inputs, and challenges associated with the existing HCM methods. The section is organized by HCM chapter, corresponding to the type of roadway or off-street facility being analyzed. Signalized Intersections Chapter 19 of the HCM 6th Edition (TRB 2016) presents methods for estimating corner circulation area, crosswalk circulation area, pedestrian delay, and pedestrian LOS score at signalized intersections. Corner Circulation Area Corner circulation area measures the average space available for pedestrians at a street corner, some of whom may be waiting to cross a street, some of whom may be actively crossing the other street, and others who may be turning the corner as they walk from one street to the other. Although this measure could be used to evaluate existing conditions, it works well for design applications, identifying the space required to serve a particular pedestrian demand at a desired QOS. The measure had LOS letters associated with it in the 1985 and 2000 editions of the HCM. However, because the pedestrian LOS score was designated as the sole pedestrian service measure for signalized intersections starting with the HCM 2010, only qualitative descriptions of service quality are now provided (in Exhibit 19-28). The ranges of pedestrian space values provided in the exhibit appear to be in error, being values used for sidewalks, walkways, and crosswalks rather than queuing areas. The method was originally developed by Fruin and Benz (1984). The following inputs are required to determine corner circulation area:  Widths of both sidewalks  Corner radius

193  Directional pedestrian volumes on both crosswalks  Traffic signal cycle length  Timing information for the pedestrian phases for both crosswalks Crosswalk Circulation Area Corner circulation area measures the average space available for pedestrians crossing a street in one direction of travel. This measure can be used to evaluate existing conditions or as part of a design application, identifying the crosswalk with required to serve a particular pedestrian demand at a desired QOS. The measure had LOS letters associated with it in the 1985 and 2000 editions of the HCM. However, because the pedestrian LOS score was designated as the sole pedestrian service measure for signalized intersections starting with the HCM 2010, only qualitative descriptions of service quality are now provided (in Exhibit 19-28). The calculation of crosswalk circulation area needs to be repeated twice—one for each direction of travel—to obtain a full picture of crosswalk performance. The method was originally developed by Fruin and Benz (1984). The version of the method used in the 1985 HCM omitted the corner radius from the crosswalk area and omitted a 3-second deduction of pedestrian start-up time from the time component of time–space. FHWA research to update the Pedestrians chapter for the HCM2000 (Rouphail et al. 1998) adjusted the time component of time–space to (a) include the effective WALK and flashing DON’T WALK time (to account for pedestrian use of a portion of the flashing don’t walk interval), (b) reduce the available time by the time required for an average pedestrian to traverse the crosswalk, and (c) incorporate a platoon-dispersal factor, all based on research by Virkler et al. (1995). The time component of time–space was modified yet again in the HCM 2010 to (a) combine the effective WALK and flashing DON’T WALK into a single effective pedestrian green time and, for reasons that are not well documented, (b) remove the component related to the time required for an average pedestrian to traverse the crosswalk. The following inputs are required to determine crosswalk circulation area:  Crosswalk length.  Crosswalk effective width. According to the HCM 6th Edition (TRB 2016), “Unless there is a known width constraint, the crosswalk’s effective width should be the same as its physical width. A width constraint may be found when vehicles are observed to encroach regularly into the crosswalk area or when an obstruction in the median (e.g., a signal pole or reduced-width cut in the median curb) narrows the walking space.”  Average directional pedestrian volumes per cycle.  Traffic signal cycle length.  Timing information for the pedestrian phase.  Average pedestrian speed.  Traffic volumes conflicting with pedestrian movements: permitted left turn, right turn, and right- turn-on-red. These volumes reduce the available crosswalk time–space by an assumed area used by vehicles encroaching on the crosswalk, a process described as “crude” during the development of the HCM2000 (Rouphail et al. 1998), but which nevertheless remains a part of the method. Research conducted by the New York City DOT found that pedestrian speeds in signalized crosswalks can be reduced by as much as 1.0 ft/s when significant opposing pedestrian volume is present, but the current HCM procedure does not account for this effect (Park et al. 2014). The New York City DOT has also found that pedestrian speeds in crosswalks vary by pedestrian gender, pedestrian signal display (WALK, flashing DON’T WALK, or DON’T WALK), time of day (a.m. or p.m.), and whether the pedestrian is alone or in a group (Peters et al. 2015).

194 Pedestrian Delay Pedestrian delay measures the average wait time from the time a pedestrian arrives at a street corner to when the pedestrian is able to enter the crosswalk. The measure had LOS letters associated with it in the 1985 and 2000 editions of the HCM (A = <10 s, B = ≥10–20 s, C = >20–30 s, D = >30–40 s, E = >40–60 s, and F = >60 s). With the adoption of the pedestrian LOS score as the pedestrian service measure for signalized intersections, the HCM 2010 and HCM 6th Edition no longer present these LOS ranges. The HCM 6th Edition provides little quantitative guidance about the implications of different delay values; however, the HCM references Dunn and Pretty (1984), who found that pedestrian noncompliance with pedestrian signals is high when average delay exceeds 30 seconds. The HCM also indicates that noncompliance is low when average delay is less than 10 seconds. A study of pedestrian compliance with signals in Washington, D.C. found that delay was but one factor influencing a pedestrian’s decision to cross on red, with other factors including: size of gaps in traffic, traffic volumes, conflicts with left-turning traffic, presence of all-red interval, and whether the pedestrian had restricted mobility (Dudata et al. 2014). Rouphail et al. (1998) note that “intersections with high conflicting traffic and/or large street widths have excellent compliance, primarily because pedestrians have no choice but to wait.” The term “noncompliance” is generally meant to include noncompliance with both flashing DON’T WALK signals and DON’T WALK signals. The only inputs required by the HCM method to determine pedestrian delay are the traffic signal cycle length and timing information for the pedestrian phase. Deficiencies of the HCM delay method mentioned in the literature include the following (Zhao and Liu 2017, Furth and Wang 2015, Kittelson & Associates, Inc. 2015):  Assumption of random pedestrian arrivals. Due to the effects of upstream signals, pedestrians may not arrive randomly at the street corner. For example, the combination of block lengths and traffic signal cycle lengths used in downtown Portland, OR provides pedestrian progression in the opposite direction of vehicular traffic on one-way streets.  Applies only to single-stage crossings. Pedestrian movements at signalized intersections—and even more so at alternative intersection forms—can involve crossing more than one crosswalk or waiting on an island to continue crossing in the same direction. Pedestrian arrival patterns for all stages after the first are not random.  Does not account for pedestrian-actuated signals and beacons (e.g., HAWKs), which can result in substantially lower pedestrian delay compared to the HCM estimate. Pedestrian LOS Score The pedestrian LOS score is the measure used to determine signalized intersection pedestrian LOS (i.e., the service measure) in the HCM 2010 and HCM 6th Edition. It is also an input for determining urban street segment and facility LOS, measures of a pedestrian’s overall experience walking along a street. Unlike more traditional HCM measures such as delay, the pedestrian LOS score cannot be measured directly in the field. Instead, it is a unitless value calculated from a regression equation that estimates an average pedestrian’s satisfaction with the crossing, based on various field-measurable inputs. The score is converted into LOS letters as follows: A = ≤1.50 , B = >1.50–2.50, C = >2.50–3.50, D = >3.50–4.50, E = >4.50–5.50, and F = >5.50. The method used in the HCM 6th Edition was originally developed by Petritsch et al. (2005), based on a “Walk for Science” event held in Sarasota, FL in which 45 participants first rated videos of various signalized intersection crossings and then walked a 3-mile course, rating their satisfaction with each of 23 signalized intersection crossings. Some of the intersections seen in the video clips matched actual

195 intersections the raters would walk through later; the researchers found there was a difference between video and in-field ratings and adjusted the video ratings accordingly. The inclusion of the video clips allowed right-turn channelizing islands to be included in the model. NCHRP Project 03-70 expanded on the video lab method tested in the “Walk for Science” (Dowling et al. 2008). A total of 32 video clips were developed from filming sessions in Tampa, FL and San Francisco, CA, each depicting the view of a pedestrian walking along a street segment, followed by crossing a side street at a signalized intersection, under varying conditions. Video lab participants were recruited in New Haven, CT, Chicago, IL; Oakland, CA; and College Station, TX, where they were shown, one mode at a time, video clips depicting walking, cycling, and driving conditions, and asked to rate the conditions shown in each clip on an A-F scale, based on how satisfied they were as a traveler. For the pedestrian testing, four clips were shown in every city, along with six additional clips only shown one of the four cities. Between 34 and 39 individuals participated in the video labs in each city. Ranges of conditions depicted in the videos included:  Traffic speed: 20–50 mph  Lanes crossed at the intersection: 2–9  Signal delay at the intersection: 0–98 s  Traffic flow rates in the outside lane: 0–939 veh/h  Pedestrian volumes: low, medium, high (volumes not specified) NCHRP Project 03-70 (Dowling et al. 2008) tested a number of different model forms, but ended up recommending an urban street segment model that combined the Petrisch et al. (2005) signalized intersection LOS model with the urban street link (i.e., between intersection) pedestrian LOS model developed by Landis et al. (2001). A follow-up implementation phase to NCHRP Project 03-70 tested the proposed multimodal models at workshops with volunteer agencies in a number of regions around the U.S. The NCHRP project team conducted workshops with staff from Arlington County, VA; Atlanta (GA) Regional Commission; City of Boise and Ada County Highway District, ID; Delaware Valley Regional Planning Commission, Philadelphia, PA; Cities of Portland, Hillsboro, and Gresham, OR; San Antonio– Bexar County MPO (TX); and City of San Diego, CA. The Florida DOT sponsored an additional three workshops with state DOT, MPO, and city staff in Tallahassee, Gainesville, and Tampa, FL. At each workshop, staff tested all the multimodal methods on arterial street facilities they were familiar with. No changes to the signalized intersection method (other than correcting two typos in the equation) were identified as being needed as a result of the testing (Dowling et al. 2010). The HCM 6th Edition (TRB 2016) identifies locations with a “free (i.e., uncontrolled) channelized right turn with multiple lanes or high-speed operation” as a limitation of the method, as this condition was not included in the video testing. Sensitivity of the Pedestrian LOS Score This section evaluates the sensitivity of the signalized intersection pedestrian LOS score to its input factors, by applying the HCM’s pedestrian LOS methods. The sensitivity of a given input factor is assigned to one of three categories:  Insensitive. Pedestrian LOS is unlikely to change when the factor is varied over its reasonable range.  Moderately sensitive. Pedestrian LOS may change by up to one letter when the factor is varied over its reasonable range.  Highly sensitive. Pedestrian LOS may change by more than one letter when the factor is varied over its reasonable range. The pedestrian LOS score incorporates the following input factors:

196  Number of travel lanes being crossed (including turn lanes)  Number of right-turn channelizing islands encountered on the crossing  Traffic volume per 15 minutes traveling over the crosswalk  Right-turn-on-red volume per 15 minutes conflicting with pedestrian movements  Permitted left-turn volume per 15 minutes conflicting with pedestrian movements  Midblock 85th percentile traffic speed on the street being crossed  Pedestrian delay To test sensitivity, a base intersection was defined with the following attributes:  Number of travel lanes crossed: 5  Number of right-turn channelizing islands: 0  Outside lane traffic flow rate: 500 veh/h  Right-turn-on-red volume per 15 minutes: 10 veh  Permitted left-turn volume per 15 minutes: 40 veh  Midblock traffic speed: 30 mph  Pedestrian delay: 35 sec (comparable to 90-second cycle length with 10-second effective pedestrian walk time) One attribute was varied at a time to determine the resulting pedestrian LOS score, which is used to determine pedestrian LOS. Figure A1 shows the results. It can be seen that pedestrian LOS at signalized intersection is  Insensitive to pedestrian delay, right-turn-on-red volume, and permitted left-turn volume;  Moderately sensitive to street width and speed; and  Highly sensitive to channelizing island presence and traffic volume. Simplified Versions of the HCM Method The Planning and Preliminary Engineering Applications Guide to the HCM (PPEAG) (Dowling et al. 2016) refers readers to the HCM equations for calculating the signalized intersection pedestrian LOS score, but provides a larger set of suggested default values than the HCM does, with the result that the only inputs the user must supply are the outside lane traffic flow rate and the number of lanes being crossed. The Florida Q/LOS Handbook (FDOT 2018) only considers conditions between intersections in evaluating a planning-level pedestrian LOS, and not conditions at a signalized intersection. The HCM 6th Edition, in Chapter 18: Urban Street Segments, also suggests evaluating conditions only between intersections (a “link- based” evaluation) as a means of reducing the amount of required data for a pedestrian analysis, particularly when larger study areas are involved.

197 (a) Delay (b) Street width (traffic lanes) (c) Speed (d) Right-turns-on-red and permitted left turns (e) Channelizing Island Presence (f) Outside lane flow rate Figure A1. Sensitivity of HCM Pedestrian Signalized Intersection LOS to Input Factors. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 20 40 60 80 100 120 140 In te rs e ct io n P e d e st ri an L O S Sc o re Delay (seconds) LOS F LOS E LOS D LOS C LOS B LOS A 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 1 2 3 4 5 6 7 8 9 In te rs e ct io n P e d e st ri an L O S Sc o re Number of Traffic Lanes Crossed LOS F LOS E LOS D LOS C LOS B LOS A 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 10 20 30 40 50 In te rs ec ti o n P ed es tr ia n L O S Sc o re 85th Percentile Speed of Street Being Crossed (mph) LOS F LOS E LOS D LOS C LOS B LOS A 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 10 20 30 40 50 60 In te rs e ct io n P e d e st ri an L O S Sc o re Right Turn on Red Volume (veh/15 min) LOS F LOS E LOS D LOS C LOS B LOS A 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 1 2 In te rs e ct io n P e d e st ri an L O S Sc o re Number of Channelizing Islands on Crossing LOS F LOS E LOS D LOS C LOS B LOS A

198 Urban Street Segments Chapter 18 of the HCM 6th Edition (TRB 2016) presents methods for estimating pedestrian LOS for an urban street segment, a street section bounded by intersections where motorized vehicle traffic may have to stop. However, the pedestrian LOS method can only be applied to street segments where both boundary intersections are signalized, as no pedestrian LOS methods have been developed for roundabouts or all- way stops. In addition, at the time the HCM 6th Edition was published, no research had been performed to determine whether pedestrian satisfaction at signalized interchange ramp terminals or alternative intersections was different from other forms of signalized intersections. An urban street segment pedestrian LOS score is calculated from three components:  Urban street pedestrian link (i.e., between signalized intersection) LOS score  Signalized intersection pedestrian LOS score  Midblock crossing difficulty The LOS letter obtained from the segment pedestrian LOS score is then compared to the LOS letter obtained from a separate evaluation of pedestrian density on the pedestrian facility, with the worse of the two LOS letters (traveler perception or operational) used as the pedestrian LOS for the urban street segment. These calculations are performed for the pedestrian facilities on both sides of the street, even when a sidewalk is not provided, or a sidewalk is provided on one side of the street but not the other. Pedestrian LOS Score The method used in the HCM 6th Edition to determine a pedestrian link LOS score was originally developed by Landis et al. (2001), based on a “Walk for Science” event held in Pensacola, FL in which 75 participants walked a 5-mile course, rating their satisfaction with the conditions experienced along each segment. A simple regression model was developed to predict LOS, based on the participant responses to the varying conditions they experienced during their walk. As described above for signalized intersections, NCHRP Project 03-70 (Dowling et al. 2008) used video labs in developing models of pedestrian LOS, in which participants rated the walking environment shown in video clips, each of which depicted a walk along a street section, followed by crossing a side-street at a signalized intersection. A number of different model forms were tested, and neither unsignalized conflict frequency (i.e., intersections and driveways) nor pedestrian density were found to be significant in determining pedestrian satisfaction, with the caveat that the range of pedestrian density shown in the clips was relatively low. Dowling et al. (2008) proposed two candidate models for determining urban street segment pedestrian LOS, both of which incorporated the Petrisch et al. (2005) signalized intersection pedestrian LOS model and the Landis et al. (2001) urban street link pedestrian LOS model. One model provided a better statistical fit with the video lab data, but did not produce LOS F for any of the video clips. The other model was developed by manually adjusting the parameters of the first model to produce a full range of A-F results from the scenes shown in the video clips. The first model ended up being used in the HCM. In addition, at the request of the panel, the researchers developed a “roadway crossing difficulty” factor that adjusts the segment pedestrian LOS score up or down based on the difficulty of making a midblock crossing. This factor compares delay (walking plus waiting) diverting to one of the boundary signalized intersections, with delay (walking plus waiting) making a crossing at a legal midblock location. The assumed starting point for calculating average diversion delay is one-third the distance along the link. As described above for signalized intersections, a follow-up implementation phase to NCHRP Project 03-70 tested the urban street pedestrian LOS method at workshops in 10 regions around the U.S (Dowling

199 et al. 2010). As a result of this testing, which included conditions not present in the model development, the following modifications were made to the link component of the model:  The barrier effect of on-street parking was increased.  The assumed horizontal separation between pedestrians and motor vehicles was increased in situations where shoulder, parking lane, or bicycle lane striping existed.  The positive effect of additional sidewalk width was capped at 10 feet, as the original model showed an illogical decreasing trend for widths greater than 10 feet.  The effect of motor vehicle volumes on pedestrian LOS on low-volume roadways without sidewalks was reduced. The HCM 6th Edition created separate LOS thresholds for pedestrian link and segment scores. This was done in conjunction with adding guidance on using the link score by itself, so that the link thresholds matched those used in the original Landis et al. (2001) research. The segment model was designed so that the pedestrian, bicycle, transit, and auto segment models would produce scores using the same set of thresholds. The link thresholds are as follows: A = ≤1.50 , B = >1.50–2.50, C = >2.50–3.50, D = >3.50– 4.50, E = >4.50–5.50, and F = >5.50. The segment thresholds are as follows: A = ≤2.00 , B = >2.00–2.75, C = >2.75–3.50, D = >3.50–4.25, E = >4.25–5.00, and F = >5.00. The HCM 6th Edition changed the weighting of the link and intersection LOS scores used in determining segment LOS. The HCM 2010 used a distance-based weighting (i.e., link length and intersection crossing length); the HCM 6th Edition changed this to a time-based weighting (i.e., time to walk the length at the average pedestrian speed and intersection delay) based on committee input, as a time-based weighting was thought to better represent the relative exposure of pedestrians to both elements of the segment. Sensitivity of the Pedestrian Link LOS Score This section evaluates the sensitivity of the pedestrian link LOS score to its input factors, by applying the HCM’s pedestrian LOS methods. The sensitivity of a given input factor is assigned to one of three categories:  Insensitive. Pedestrian LOS is unlikely to change when the factor is varied over its reasonable range.  Moderately sensitive. Pedestrian LOS may change by up to one letter when the factor is varied over its reasonable range.  Highly sensitive. Pedestrian LOS may change by more than one letter when the factor is varied over its reasonable range. The pedestrian link LOS score incorporates the following input factors:  Outside through travel lane width  Parking lane/shoulder width  Bicycle lane width  Percent on-street parking occupied  Buffer width between roadway and sidewalk  Sidewalk width  Outside lane flow rate  Motor vehicle running speed  Street tree/barrier presence

200 To test sensitivity, a base street was defined with the following attributes:  Outside through travel lane width: 12 feet  Parking lane width: 0 feet  Bicycle lane width: 0 feet  Percent on-street parking occupied: 0%  Buffer width: 0 feet  Sidewalk width: 6 feet  Outside lane flow rate: 500 veh/h  Motor vehicle running speed: 30 mph  Street trees: no One attribute was varied at a time to determine the resulting pedestrian link LOS score, which can be used to determine a pedestrian LOS letter for the link. Figure A2 shows the results. (a) Traffic volume (b) Traffic speed (c) Lane widths and parking (d) Sidewalk and buffer width Figure A2. Sensitivity of HCM Pedestrian Link LOS to Input Factors.

201 It can be seen that pedestrian LOS for links is  Relatively insensitive to roadway lane widths, traffic speed, landscape buffer width, and sidewalk width (except when installing a sidewalk where none existed previously);  Moderately sensitive to the percentage of occupied on-street parking and the presence of street trees; and  Highly sensitive to traffic volume. Sensitivity of the Pedestrian Segment LOS Score The pedestrian segment LOS score weights the intersection and link scores equally. Therefore, if the average time required to walk the length of the link equals the average intersection delay, the segment LOS score will be the average of the intersection and link scores. If delay exceeds walking time, the intersection LOS will have a greater influence on segment LOS, while if walking time exceeds delay, the link LOS will have a greater influence on segment LOS. The roadway crossing difficulty factor includes three components: (1) the lowest of “diverting to the nearest signalized intersection delay”, “diverting to a midblock crossing delay”, and 60 seconds; (2) the pedestrian link LOS score; and (3) the pedestrian intersection LOS score. When computing the segment LOS score, the link score is multiplied by this factor, which is constrained to the range of 0.80–1.20. Figure A3 shows the sensitivity of the adjusted link LOS score to each of these inputs, assuming the initial link and intersection pedestrian LOS scores shown in the figure. Figure A3. Sensitivity of the Roadway Crossing Difficulty Factor to Input Factors. A change in delay (for example, by installing a midblock crossing) typically only affects the factor in a relatively narrow range of 30–40 seconds of delay, after which no further effect is seen, due to the upper and lower constraints on the value of the factor. The factor has a greater effect on links with poor pedestrian facilities, where the initial score can be adjusted by as much as ±1.0 point when the starting link LOS is E (a score of 5). In comparison, the maximum adjustment is ±0.2 point when the starting link LOS is A (a score of 1). All other things being equal, a better intersection LOS will result in a better segment LOS in

202 the middle of the possible range of delays (approximately 20–50 seconds), but the same segment LOS when delays are very low, or are greater than or equal to 60 seconds. The relatively low range of possible delay values (≤60 seconds) means that the diversion component of delay may use up most or all of the delay. At an average walking speed of 4.4 ft/s, a pedestrian walks 44 ft in 10 s, which equates to 132 feet of segment length because the starting point for the diversion is assumed to be one-third the segment length. Thus, even if average signal delay to cross at the boundary intersection is a very 1ow 10 seconds, traffic signals would have to be spaced more frequently than every 660 feet for diversion to possibly show any benefit. If traffic signals are spaced every 800 feet and a midblock crossing is provided halfway between the signals, diversion time from 1/3 the way along the segment to the midpoint of the segment would take slightly more than 30 seconds, meaning that crossing delay at the midblock crossing would need to be less than 30 seconds (and more likely 20 seconds) to see any change in the segment LOS score. In addition, if a pedestrian-actuated beacon were to be provided midblock, the existing HCM delay equation does not account for the reduced delay provided by pedestrian actuation and would therefore overestimate delay, reducing any possible change in the score. Consequently, although pedestrian segment LOS is mathematically highly sensitive to the roadway crossing factor, for practical purposes it is difficult to devise a crossing treatment that will produce much of an effect on the segment LOS score. Simplified Versions of the HCM Method The PPEAG (Dowling et al. 2016) uses the HCM urban street link method to determine LOS. It also provides a larger set of suggested default values than the HCM does, a computational engine for the pedestrian link method (as well as the bicycle and transit methods), and guidance on analyzing less-common situations. The PPEAG also offers guidance that pedestrian density only becomes a potential factor in determining pedestrian LOS when pedestrian volumes exceed 1,000 ped/h. The Florida Q/LOS Handbook (FDOT 2018) develops a planning-level pedestrian LOS estimate based on traffic volumes and the percentage of the facility where a sidewalk is present. The HCM 6th Edition, in Chapter 18: Urban Street Segments, offers a link-based evaluation as a means of reducing the amount of required data for a pedestrian analysis, particularly when larger study areas are involved, but notes that some factors, including intersection conditions, roadway crossing difficulty, and pedestrian density will not be incorporated in the evaluation. Reformulation of the HCM Method Ali et al. (2010) conducted a new analysis of the NCHRP Project 03-70 pedestrian video clips, using a cumulative logistic regression model, rather than the simple regression used in NCHRP Project 03-70. A data visualization analysis was first used to identify ranges of values of the independent variables that explained similar variation in the dependent variable; the following categories resulted from this analysis:  Sidewalk width: 0–5 feet, >5 feet  Outside lane width: 10–12 feet, >12 feet  On-street parking: 0%, >0%  Same-direction traffic volume: 0–500 veh/h, 501–1,500 veh/h, >1,500 veh/h  Traffic speed: 20–40 mph, >40 mph  Number of traffic lanes: 1, 2, 3, 4 The presence of absence of a barrier was found to have the second-highest correlation to the satisfaction score, after sidewalk width (Ali et al. 2010). However, the barrier variable was also highly correlated with sidewalk width, indicating that the study sites with wider sidewalks also had a barrier present. Therefore, the barrier variable was not included in the remainder of the modeling effort.

203 The final model requires less data collection and is capable of predicting the probability of a user selecting each LOS letter for a given set of conditions. Testing of the model indicating it had a tendency to underpredict the number of LOS A ratings (Ali et al. 2010). Pedestrian Space Average pedestrian space on a pedestrian facility reflects pedestrians’ ability to (1) maintain their desired walking speed and (2) freely chose their path across the path of other pedestrians (TRB 1980). The method used in the HCM 6th Edition dates back to Transportation Research Circular 212 (TRB 1980), which drew largely from work by Fruin (1971) and Pushkarev and Zupan (1975). The LOS ranges proposed in Circular 212, and eventually incorporated into the 1985 HCM were developed from a review of three different researchers’ categorizations of pedestrian space (Oeding 1963, Fruin 1971, Pushkarev and Zupan 1975). There was good agreement between the researchers on the middle part of the LOS range, so the development of an LOS recommendation focused on defining the upper (LOS A/B) and lower (LOS E/F) ends of the range. Circular 212 also recognized that, unlike motorized vehicle traffic flow, pedestrian flows could be quite variable over short spans of time. In particular, the combination of short blocks and frequent signalized intersections found in the central business districts of many large cities tended to create groups of pedestrians (platoons) in which it was difficult to choose one’s desired speed and path: “It is clear that an ‘average’ flow rate, even if refers to a period as short as one minute, may not be entirely relevant to defining the condition of the majority of pedestrians in a traffic stream who are in platoons. To the pedestrian within a platoon, it is small consolation that a few seconds prior to his arrival, the section of walkway on which he is now experiencing congested conditions was virtually empty. For example, if the objective is to provide a relatively high measure of mobility, then the time period truly relevant for design does not appear to be 15 minutes, 1 minute, or any other arbitrary time span, but rather the intermittent periods during which flow in platoons occurs.” (TRB 1980). Pushkarev and Zupan (1975) had found that the upper limit of platoon flow was approximately 4 ped/ft/min higher than average flow. Using this relationship, Circular 212 determined that the LOS for platoon flow was approximately one letter worse than for average flow, for a given value of average pedestrian space. Circular 212 recommended setting LOS thresholds at the following pedestrian space values (in ft2/ped): 130, 40, 24, 16, 11, and 6, with the LOS A/B threshold being 130 ft2/ped for platoon flow and 40 ft2/ped for average flow, and the LOS E/F threshold being 11 ft2/ped for platoon flow and 6 ft2/ped for average flow (TRB 1980). The 1985 HCM presented a single table for average conditions with thresholds of 130, 40, 24, 15, and 6 ft2/ped, along with text indicating that the LOS for platoon flow “is generally one level poorer than that determined by average flow criteria, except for some cases of LOS E, which encompasses a broad range of pedestrian flow rates” (TRB 1985). The FHWA research to develop the Pedestrians chapter for the HCM2000 (Rouphail et al. 1998) revisited the pedestrian LOS criteria and recommended that separate LOS thresholds be developed for average and platoon flow, similar to what Circular 212 (TRB 1980) had recommended. For platoon flow, Rouphail et al. (1998) recommended setting the LOS A/B boundary at the lowest flow rate at which Pushkarev and Zupan (1975) had observed platoon flow, and setting the LOS E/F boundary at Pushkarev and Zupan’s pedestrian flow rate at capacity. The recommended pedestrian space thresholds of 530, 90, 40, 23, and 11 ft2/ped were subsequently incorporated into the HCM2000 and have remained in place ever since. In addition, Rouphail et al. (1998) recommended lowering the LOS A/B threshold for average flow to 60 ft2/ped, based on input from the TRB Highway Capacity Committee, and raising the LOS E/F threshold to 8 ft2/ped, based on a review of the pedestrian flow literature. These changes were also incorporated into the HCM2000 and retained in subsequent editions. One final change was made with the HCM 2010, where density-based (i.e., pedestrian space) LOS was evaluated alongside perception-based

204 (i.e., pedestrian segment LOS score) LOS, with the lower of the two LOS letters being reported as the LOS result for the urban street segment. Determining average pedestrian space requires the following inputs:  Pedestrian free-flow speed (the HCM recommends a value of 4.4 ft/s, except when more than 20% of the pedestrians are age 65 and older, in which case a value of 3.3 ft/s is recommended. Free-flow speed is reduced by 0.3 ft/s if the sidewalk grade is +10% or greater.  Effective sidewalk width, the actual sidewalk width minus adjustments for shy distance that pedestrians give objects in (e.g., lamp posts) or adjacent to (e.g., building faces) the walkway. A review of the HCM method by the New York City Department of City Planning (2006) identified the following weaknesses of the method:  Bi-directional or multi-directional flow is not accounted for in the method; friction from opposing pedestrians could affect pedestrian flow.  No source in the literature, including Pushkarev and Zupan (1975) describe how to measure a shy distance in the field; it may be possible that shy distance varies with pedestrian volume, time of day, and/or adjacent land use.  The method was felt to be too insensitive to changes in pedestrian volume or effective width. The New York City Department of City Planning study (2006) recommended further research to address the first two issues, and tested a potential method using video cameras for measuring shy distances. However, no further study was apparently carried out. The study also found that “the number of impeded pedestrians observed at a location was an excellent predictor of pedestrian speed and subjective interpretations of the sidewalk’s LOS.” Circular 212 (TRB 1980) looked at the effect of bi-directional flow on pedestrian operations. When pedestrian volumes in each direction were roughly equal, each group occupied its own proportionate share of the sidewalk, with the result that there was little reduction in sidewalk capacity. With a 90/10 directional split, though, the minor flow occupies a greater proportion of the sidewalk area relative to its volume, a capacity reduction of 15% occurs. This reduction, however, was not incorporated into the recommended method. Sensitivity of Pedestrian Space LOS Figure A4 illustrates the sensitivity of pedestrian space LOS to changes in pedestrian flow rates and effective sidewalk widths, for both average and platoon flow. For a given sidewalk operating with platoon flow at the LOS C/D threshold, pedestrian flow rates would need to increase by 64% or the effective sidewalk width would need to decrease by 39% for the LOS to degrade by one letter. Similarly, the pedestrian flow rate would need to decrease by 32% or the sidewalk width would need to increase by 47% for the LOS to improve by one letter. For example, for a 6-ft sidewalk with a pedestrian flow rate of 2,300 ped/h, the flow rate would need to increase to 3,775 ped/h or the effective sidewalk width would need to shrink to 3.6 ft for the LOS to drop by one letter, which indicates that pedestrian space LOS is relatively insensitive to its input factors. Under platoon flow, the LOS B/C boundary is approximately 700 ped/h for a 4-ft effective sidewalk width, 1,050 ped/h for a 6-ft width, and 1,750 ped/h for a 10-ft width; these values more than double when average flow is considered. As a result, most pedestrian facilities have no need for a pedestrian space assessment.

205 (a) Average flow (b) Platoon flow Figure A4. Sensitivity of Pedestrian Space LOS to Input Factors. Simplified Versions of the HCM Method Both the PPEAG (Dowling et al. 2016) and the HCM 6th Edition note that pedestrian space is unlikely to control the overall urban street segment LOS when pedestrian volumes are less than 1,000 ped/h. Given the simplicity of the full HCM method, no simpler methods are presented in either document. Urban Street Facilities Chapter 16 of the HCM 6th Edition (TRB 2016) presents a method for estimating pedestrian LOS for an urban street facility, defined as a series of consecutive urban street segments. The pedestrian facility LOS score is calculated as the time-weighted average of the individual segment LOS scores, with “time” consisting of the sum of the walking time and the downstream boundary intersection delay for a given segment. The LOS scale is the same as that used for urban street segments. The method was developed by Dowling et al. (2006) as part of an overall multimodal urban street LOS analysis framework. The HCM 6th Edition changed the weighting used to calculate facility LOS from distance-based to time-based. The facilities method requires all the input data and is subject to all the sensitivities associated with the link, signalized intersection, and segment methods. Two-Way Stop-Controlled Intersections and Unsignalized Midblock Crossings Chapter 20 of the HCM 6th Edition (TRB 2016) presents a method for estimating pedestrian LOS based on average pedestrian delay when crossing the major (i.e., uncontrolled) street at a TWSC intersection or at a marked midblock crosswalk. The method was originally developed by NCHRP Project 03-92 during the production of the HCM 2010. Pedestrian delay measures the average wait time from the time a pedestrian arrives at a street corner to when the pedestrian is able to enter the crosswalk. The delay ranges associated with each LOS are as follows: A = 0–5 s, B = 5–10 s, C = 10–20 s, D = 20–30 s, E = 30–45 s, and F = >45 s. Delay is estimated by a gap acceptance model that considers the time required for an average pedestrian to make the crossing, the average headway between vehicles on the street, and a user-specified motorist yielding rate (TRB 2010). Because the yielding model considers the probability that the first arriving vehicle will yield and, if not, that the second arriving vehicle will yield, and so on, the method is not straightforward to use by hand. A

206 computational engine for the method was developed by NCHRP Project 03-92, but has not been posted on HCM Volume 4 by decision of the TRB Highway Capacity Committee (to avoid the need to provide user support and training for this and other spreadsheet-based computational engines). The method is limited to analyzing crossing four through travel lanes in one stage. The HCM 6th Edition does not provide guidance on analyzing a street with four travel lanes and a two-way left-turn lane (TWLTL), but depending on pedestrian behavior, it could be modeled as a two-stage crossing (two lanes each) or as a wide one-stage crossing of four lanes. The HCM provides a table of yielding rates from the literature available at the time the chapter was originally developed (around 2009), for a limited selection of crossing treatments. The text accompanying the table encourages analysts to use local values when available. Sensitivity of Pedestrian Delay LOS to Input Factors This section evaluates the sensitivity of pedestrian LOS to the input factors used in calculating it. As before, sensitivity of a given input factor is assigned to one of three categories:  Insensitive. Pedestrian LOS is unlikely to change when the factor is varied over its reasonable range.  Moderately sensitive. Pedestrian LOS may change by up to one letter when the factor is varied over its reasonable range.  Highly sensitive. Pedestrian LOS may change by more than one letter when the factor is varied over its reasonable range. Estimating pedestrian delay requires the following inputs:  Average pedestrian speed (can be defaulted)  Crossing width  Two-stage crossing (yes/no)  Traffic volume  Motorist yielding rate To test sensitivity, a base street was defined with the following attributes:  Crossing width: 52 feet (e.g., a 4-lane undivided street with no parking, bicycle, or turn lanes)  One-stage crossing  Traffic volume: 500 veh/h  Motorist yielding rate: 0% One attribute was varied at a time to determine the resulting pedestrian delay and corresponding LOS. The 5-lane (TWLTL) scenario was modeled as a wide one-stage crossing of four lanes. Figure A5 shows the results. It can be seen that pedestrian LOS for uncontrolled crossings is highly sensitive to crossing width, traffic volumes, yielding rate, and number of crossing stages.

207 (a) Traffic volume, 0% yielding (b) Motorist yielding rate (c) Crossing width (d) Traffic volume, 100% yielding Figure A5. Sensitivity of HCM Pedestrian Uncontrolled Crossing Delay LOS to Input Factors. In investigating the relatively high delay associated with a motorist yielding rate of 100% (Figure A5b), a problem with the yielding portion of the methodology was discovered. Some small amount of average delay would be expected, even with a 100% yielding rate, because some pedestrians would arrive at the crosswalk with an arriving vehicle too close to the crosswalk to be able to yield. However, the yielding model does not consider this case. Instead, the method produces delay equivalent to 0% yielding up to a volume of 409 ped/h. With one additional ped/h, the delay decreases by nearly half and then continues to decrease slowly as traffic volumes increase (Figure 5d). There appear to be two causes for this illogical model behavior:  First, the variable n, the average number of crossing events before an adequate gap is available, is calculated as the gap delay for pedestrians who incur nonzero delay divided by the average vehicle headway for each through lane, rounded down to the next lower integer. When yielding rates are high, the gap delay is low and n calculates as zero. This does not seem to be an intended result, because n is used in a summation of crossing event probabilities that runs from 1 to n.  Second, if one forces n to have a minimum value of 1, the calculated delay at low volumes with 100% yielding is considerably higher than the calculated delay with 0% yielding. This appears to be a result of the equation term for calculating delay when motorists yield, which is half the

208 vehicular headway for the first yielding event. The vehicular headway decreases as vehicular volumes increase, which would account for the shape of the curve. As a result, even though the current HCM uncontrolled crossing delay model is highly sensitive to its input factors, and can be affected by many types of pedestrian safety countermeasures that influence the driver yielding rate, the theory behind the model needs to revisited to address the illogical model behavior. In addition, the model was not field-tested as part of its development during NCHRP Project 03-92 (Production of the HCM 2010). Simplified Versions of the HCM Method Neither the HCM 6th Edition (TRB 2016) nor the PPEAG (Dowling et al. 2016) provide a simplified version of the method. Applications and Evaluations of HCM Methods in the Literature The city of Geneva, NY employed the HCM 6th Edition PLOS methodology in its Active Transportation Plan (Barton & Loguidice, DPC and Sprinkle Consulting, Inc. 2017). The consultants who developed the plan collected primary data on the presence and width of segment sidewalks, the width of outside travel lanes, traffic volumes and speeds, the presence and width of buffers, and barriers (e.g., on-street parking, and street trees). The plan provides a color-coded network map of pedestrian segments, though not intersections. It also gives visual examples of lettered PLOS scores for efficient discernment of segments with high and low PLOS. The plan complements PLOS scoring and visualizations with maps of roadway topography (i.e., slope), which influences pedestrians’ experience of comfort. Notably, the plan draws upon the analysis’ HCM PLOS results to prioritize recommendations on filling gaps in the city’s sidewalk network. In a review of several different pedestrian and bicycle LOS and “comfort” analytic tools including “level of stress” methodologies, WalkScore and BikeScore algorithms, and the HCM 6th Edition PLOS methods, Zuniga-Garcia, Ross, and Machemehl (2018) concluded that level of stress methods are suitable for higher- level plans, such as corridor and transportation network plans, but harbor limitations at more granular levels of analysis, such as intersections, segments, and individual facilities. They also argue that multimodal LOS and HCM mode-specific methodologies are suitable for corridor evaluations, yet both require a significant amount of user training and data collection. Additionally, Zuniga-Garcia, Ross, and Machemehl (2018) explore use of the PEQI developed by the San Francisco Department of Public Health to complement the HCM, which incorporates measures of “perceived pedestrian safety,” such as pedestrian-scale lighting. The Oregon DOT (2017) has developed a PLTS method that is intended to be a companion to the Bicycle Level of Traffic Stress method (Mekuria et al. 2012), for use as the preferred method for long-range transportation plans and as a screening tool for more-detailed studies with smaller study areas (i.e., interchange-area plans, facility plans, project development, development review). The following input data are required, many of which are routinely collected:  Roadway segments:  Sidewalk width  Sidewalk condition  Buffer type and width  Bicycle lane width  Parking width  Number of travel lanes  Posted speed

209  Illumination presence  General land use  Roadway crossings:  Functional class  Number of travel lanes  Posted speed  Roadway average daily traffic (optional)  Sidewalk ramp presence  Median refuge presence  Illumination presence  General signalized intersection features Four PLTS are defined (Oregon DOT 2017):  PLTS 1: Suitable for all users, including children 10 years or younger, groups of people and people using a wheeled mobility device.  PLTS 2: Suitable for children over 10, teens and adults; some factors may limit people using wheeled mobility devices.  PLTS 3: An able-bodied adult would feel uncomfortable but safe using the facility; small portions of the facility may be impassible to people using wheeled mobility devices or require them to travel in the roadway.  PLTS 4: High traffic stress; only able-bodied adults with limited route choices would use the facility; includes roadways with no sidewalk. Sahani, Devi, and Bhuyan (2017) focus on pedestrians’ experience of delay to conceptualize signalized intersections’ PLOS provision. To estimate PLOS at 45 signalized intersections in eight mid-size Indian cities, the authors use: (1) number of lanes crossed; (2) 85th percentile speed of motorized vehicles; (3) number of left-turning vehicles per 15-minute interval; (4) number of permissible right-turning vehicles per 15-minute interval; (5) number of vehicle through movements per 15-minute interval; (6) number of left- turning nonmotorized vehicles per 15-minute interval; (7) number of permissible right-turning nonmotorized vehicles per 15-minute interval; and (8) number of nonmotorized vehicle through movements per 15-minute interval. They calculate waiting time delay and pedestrian–vehicle interaction delay during crossing in defining “delay.” The authors use ridge regression analysis to minimize the mean square error of the coefficients and introduce a penalty term in the PLOS models. They then classify PLOS scores into LOS categories A–F by employing a clustering technique to derive PLOS scores across the study intersections. Next, they compare estimates from their proposed pedestrian delay model with other models (i.e., HCM 2010; Braun and Roddin, 1978; Marisamynathan and Vedagiri delay model; deterministic queuing model; and a steady-state stochastic delay model) to determine their accuracy in predicting pedestrians’ qualitative ratings of perceived LOS. They report that across a sample of 10 sites, their proposed delay model matched the perceived LOS 90% of the time compared to HCM 2010’s 10% match. The authors find that total pedestrian delay maintained a strong, positive linear relationship with increases in waiting time delay. Further, they note how perceptions of delay are more favorable when wait times are less than 20 seconds and enter PLOS F territory when wait times exceed 70 seconds. MacDonald et al. (2018) propose considering adjacent segments (e.g., 6 in a gridded urban street network) and intersections (i.e., the ones on either end of a segment) when rating pedestrian LOS. For example, a segment with PLOS A adjacent to segments and intersections with PLOS D arguably does not afford a pleasant pedestrian experience.

210 In a systematic review of 58 studies on PLOS, Raad and Burke (2018) make several observations. First, they note that nearly half (46.4%) of the studies were published in the Transportation Research Record and that most studies applied PLOS assessments to street areas and footpaths, whereas fewer studies involved intersections or midblock crossings. A majority (70%) of the studies employed regression models, whereas 3% used fuzzy neural network approaches, and only 2% used simulation, with the remaining 25% incorporated point systems using weighting factors based upon the model developers’ perceptions of the importance of various environmental factors on the quality of walking experiences. Looking across methods, they identify groups who use modeling and precise geometric measures in PLOS estimates as “Geometricians” and those who assign points to various factors in the pedestrian environment as “Experientialists.” They note that while the number of studies on measuring PLOS has increased in recent years, so too has the level of disagreement among the Geometricians and the Experientialists. Raad and Burke ask “How do we best marry the two approaches?” and “What factors should be kept, and which discarded?” The authors argue that too many studies rely on scholars’ “best guess” as to which factors matter to higher-quality pedestrian environments. They cite the limited reporting on PLOS measures’ interrater reliability and their low levels of external validity (to what extent are PLOS estimates transferable to other urban contexts?). Raad and Burke believe it is preferable to draw upon expert opinions as well as pedestrian opinions to identify those factors worth including in PLOS models. Challenges and Limitations of HCM Methods Point–Segment–Facility Framework The HCM 6th Edition’s methodological chapters (TRB 2016) are organized by points (intersections), segments (urban street sections between boundary intersections, such as signalized intersections, including conditions at the downstream intersection), and facilities (multiple consecutive segments), following the HCM’s historic framework for evaluating the motorized vehicle mode. The current pedestrian methodology also defines links, street sections between boundary intersections, but not including conditions at the intersections. However, whereas roadway conditions (e.g., number of lanes, traffic volumes) for motorized vehicles may stay relatively the same between signalized intersections, pedestrian facility quality and volumes are more likely to change from block to block, particularly in suburban areas with long signal spacings. As a result, the HCM’s traditional organization of system elements may not meet the needs of pedestrian-focused analyses. Trip Purpose Trip purpose has not been a direct consideration in HCM methods. It can be considered indirectly in terms of the performance or design standard an agency sets for a street by functional class, location (e.g., downtown/suburban/rural), and/or time of day. However, pilot testing by NCHRP Project 03-70, which developed the current HCM multimodal methods, found differences in traveler perceptions between leisure and time-constrained trips (Dowling et al. 2008). As a result, that project’s researchers directed its video survey participants to rate conditions from the perspective of a time-constrained trip, as it best reflected conditions occurring during a typical HCM analysis (e.g., peak-period travel). A study conducted by the New York City Department of City Planning (2006) found midday non-impeded pedestrian speeds were lower than a.m. peak-period non-impeded speeds, a result it attributed to both trip purpose and a greater propensity for pedestrians to travel in groups at midday (e.g., tourists, groups of workers going to lunch). Arterial and Collector Street Focus for Urban Streets Arterials and collectors are the urban streets most likely to have auto capacity and delay problems, while local streets are not intended to provide high-speed operations. As a result, the HCM has historically only

211 provided methods for arterial and collector streets. The NCHRP Project 03-70 panel and researchers accepted the premise that the project’s multimodal methods should only address urban arterials and collectors (Dowling et al. 2008). This decision may have unintentionally constrained the resulting multimodal methods. For example, NCHRP 03-70’s recruited video raters were not particularly satisfied with cycling conditions on the collector and arterial streets they viewed, no matter how wide a bicycle lane was provided. As a result, it is difficult to use the HCM’s bicycle methods to predict bicycle LOS A or B on-street segments, which might not have been the case if scenes from local streets had been included in the rating process. Quality of Service Factors Not Incorporated by the HCM’s Pedestrian LOS Methods The HCM’s description of pedestrian QOS (TRB 2016) includes an extensive range of factors that extend well beyond the current factors incorporated into any of the HCM’s pedestrian methods. Factors identified by the HCM but not currently addressed by its methods are shown below in bold: “Environmental factors contribute to the walking experience and, therefore, to the QOS perceived by pedestrians. These factors include the comfort, convenience, safety, and security of the walkway system. Comfort factors include weather protection; proximity, volume, and speed of motor vehicle traffic; pathway surface; and pedestrian amenities. Convenience factors include walking distances, intersection delays, pathway directness, grades, sidewalk ramps, wayfinding signage and maps, and other features making pedestrian travel easy and uncomplicated. “Safety is provided by separating pedestrians from vehicular traffic both horizontally, by using pedestrian zones and other vehicle-free areas, and vertically, by using overpasses and underpasses. Traffic control devices such as pedestrian signals can provide time separation of pedestrian and vehicular traffic, which improves pedestrian safety. Security features include lighting, open lines of sight, and the degree and type of street activity.” Limitations of Current HCM Pedestrian Methods Table A11 identifies a selection of known limitations of current HCM pedestrian methods. Most of these limitations are identified in the HCM itself (TRB 2016); others come from the literature, the source research for the HCM method, or work performed by the research team (Dowling et al. 2008; Park et al. 2014; Kittelson & Associates, Inc. 2015; Kittelson & Associates, Inc. 2017). Other limitations identified in the literature, such as the fact that the sidewalk width on urban streets only affects pedestrian LOS up to a width of 10 feet (Huff and Liggett 2014), are not so much limitations as outcomes of the modeling effort (i.e., no further improvement in pedestrian satisfaction was identified beyond 10 feet) and therefore are not listed in the table.

212 Table A11. Limitations of Current HCM Pedestrian Methods. Facility Type Limitations/Needs Two-lane rural highways Urban street pedestrian LOS method not developed for:  Truly rural highway sections (e.g., walking on the highway shoulder)  Pedestrian facilities in rural communities that have no intersections where highway traffic must potentially stop/yield for other vehicles  Shared-use paths parallel to and within 35 feet of the roadway Urban streets  Cannot calculate segment/facility LOS when one or more segments do not have a signalized intersection as the boundary intersection  LOS method does not account for unsignalized midsegment crosswalks (including intersections where only the side street is stop-controlled)  LOS method does not address grades >2%  LOS method does not address pedestrian under/overcrossings  Methods do not address midblock locations with high vehicular volumes crossing the sidewalk (e.g., parking facility entrances/exits)  Midblock locations with high pedestrian crossing/entering volumes only considered in the context of capacity (maximum flow rate)  LOS method does not address unpaved or uneven pedestrian facilities  Flow method does not directly identify conditions under which pedestrians will start spilling out of the normal pedestrian circulation area  Bounds identified for the midblock crossing difficulty factor tend to improve LOS on very bad facilities and worsen LOS on very good facilities, regardless of the crossing quality  LOS method does not address shared-use paths parallel to and within 35 feet of the roadway  LOS method does not address the impact of short gaps in sidewalk (under 100 feet). Until such a methodology becomes available, short gaps may be neglected in the PLOS calculation. However, the analyst should report the fact that there are gaps in the sidewalk in addition to reporting the LOS grade. Segments with relatively long gaps (over 100 feet) in the sidewalk should be split into sub-segments and the PLOS for each evaluated separately. Signalized intersections  LOS method does not account for free-flowing right-turn lanes with multiple lanes and/or high speeds  Delay method for 2- (catty-corner) or 3-stage (closed crosswalk) crossings assumes random pedestrian arrivals to begin the 2nd and 3rd stages  Delay method does not account for actuated midblock pedestrian crossings or HAWKs  LOS method does not account for grade-separated pedestrian crossings  LOS method does not address railroad crossings with frequent train traffic  Crosswalk pedestrian speed is an input, rather than a calculation result TWSC intersections  LOS based solely on pedestrian delay (i.e., no traveler perception method)  Delay method limited to crossing four traffic lanes in a single crossing stage  Delay method does not consider platooning/gap-producing effects of upstream traffic signals  No method for evaluating pedestrian cross-flows at intersection corners  Driver yielding portion of the delay methodology not field-tested; however, yielding rates do come from the literature and can be locally adjusted  Default yielding rates provided only for five types of crossing treatments All-way stop intersections  No LOS method; limited guidance on estimating pedestrian delay Roundabouts  No LOS method; guidance provided to apply the TWSC pedestrian delay method “with care” Alternative intersections  No separate LOS method; guidance provided on which intersection forms can be analyzed using the signalized intersection methods Interchange ramp terminals  No LOS method; guidance provided on applying the signalized and TWSC delay methods to evaluate overall pedestrian delay/travel time

213 Summary Techniques for Efficient and Accurate Estimation of Pedestrian Volume and Exposure The most commonly used technologies for counting pedestrians at present are manual counts in the field and manual counts from video. When automated counts are used, passive infrared, active infrared and automatic counts from video are the most common methods (FHWA 2011, Ryus et al. 2014a). Pedestrians are particularly difficult to count because they, more than cyclists and much more than motor vehicles, do not always stay in prescribed areas and often travel in close groups that make it difficult to distinguish and count individuals and to distinguish between other road or path users. For example, the most common technology for counting pedestrians, passive infrared, is known to undercount pedestrians by 9%–19% (Schneider et al. 2009, Greene-Roesel et al. 2008) with substantial variation between different manufacturer’s devices (Ryus et al. 2016). NCHRP Web-Only Document 229 gives accuracies for common pedestrian counting technologies (Ryus et al. 2016). To distinguish between pedestrians and bicyclists on paths, a combination of technologies are often employed; for example, counting total nonmotorized users with passive infrared counters and then identifying bicycles so that path users with bicycles can be distinguished from those without (pedestrians). Other technologies (automated counts from video, thermal camera, and radio) use only one sensor to distinguish pedestrians from bicycles (Ryus et al. 2016). Challenges to collecting pedestrian data include funding and legal challenges installing counters. Costs for counting at a sufficient number of sites and for a sufficient duration, as well as managing the data, are beyond some jurisdictions’ budgets. Jurisdictions have overcome these costs by using volunteers for manual counts, crowdsourcing, justifying the counts to agencies in terms of their usefulness, and acquiring additional funding sources through grants or partnering with other agencies. To overcome quality problems, agencies conduct regular manual and automated quality assurance checks of data from automated counting devices. Three types of models are discussed in NCHRP Report 770 (Kuzmyak et al. 2014) to estimate exposure from count data: trip generation and flow models, network simulation models, and direct demand models. Direct demand modeling is the most common of these because it can be easily applied, but it is not applicable beyond the community where the data used to develop the model were obtained. This method generally requires statistical modeling software to create the model. Methods for estimating pedestrian volumes are also discussed in a recent FHWA report that produced a spreadsheet-based tool for estimating pedestrian traffic at the state and regional levels using nationally available data sources (Turner et al 2018). Performance Measures for Evaluating Pedestrian Safety, Operations, Mobility, and Satisfaction Literature has identified a number of risk factors that contribute to pedestrian crash frequency and severity. These have been classified into roadway, intersection, traffic, land use, demographics and behavior, and environmental categories. With respect to crash frequency, key pedestrian risk factors include traffic volume, pedestrian volume, measures of transit activity, land use (commercial), functional class (arterials), and presence of turn lanes (Thomas et al. 2018). Key factors that affect crash severity include older pedestrians, larger vehicles (heavy vehicles), darkness, higher speed limits including driving and impact speed, pedestrian crossing the roadway and pedestrians under the influence of alcohol (Thomas et al. 2018). Table A12 and Table A13 summarize the key factors and their effects on crash frequency and severity.

214 Table A12. Key Factors Affecting Pedestrian Crash Frequency. Factors Effect on Pedestrian Crash Frequency Traffic Volume Increase in crash frequency Pedestrian Volume Increase in crash frequency Transit Activity Increase in crash frequency Land use (Commercial) Increase in crash frequency Presence of turn lanes (right, left) Increase in crash frequency Functional class (Arterial) Increase in crash frequency Table A13. Key Factors Affecting Pedestrian Crash Severity. Factors Effect on Pedestrian Crash Frequency Older pedestrians Increase in crash severity Large vehicles (heavy trucks) Increase in crash severity Darkness Increase in crash severity Higher speed limits Increase in crash severity Pedestrians crossing the roadway Increase in crash severity Pedestrians under the influence of alcohol Increase in crash severity Pedestrian speed is a key factor while describing pedestrians operations and mobility. A number of factors influence pedestrian walking speeds including environmental, traffic, and pedestrian characteristics. These have been summarized in Table A14. Table A14. Key Factors Affecting Pedestrian Speed. Factors Effect on Pedestrian Speed Age Older pedestrians walk slower than younger pedestrians Gender Males exhibit higher speeds than females Group size Slower speeds when pedestrians cross in groups Delay Pedestrian crossing speed increases with increase in delay Gap Pedestrians who accept shorter gaps have higher walking speeds Arrival during the signal phase Pedestrians arriving during the clearance phase who choose to cross may have higher walking speeds The relationship between pedestrian speed, flow, and density has been extensively described in the literature. Most of the studies describe a linear relationship between pedestrian speed and density. However non-linear fits have been found for different trip purposes (Ishaque and Noland, 2008). Studies have also found capacity losses resulting from bi-directional flows (Navin and Wheeler 1969, Fruin 1971, Al-Masaeid et al. 1993). Pedestrian satisfaction with the built environment and effects on behavior has been extensively explored in the literature. Factors affecting level of satisfaction with the pedestrian environment include those related to physical infrastructure (sidewalk width, presence, and continuity; slope; bus shelter availability; parking; crosswalk presence; pedestrian signal presence; median island presence); road safety (traffic volume, noise

215 and fumes, pedestrian flow rate, waiting time, crossing distance); aesthetics (sidewalk cleanliness and surface quality, presence of obstructions, presence of trees); access and facilities (disabled pedestrian access, land use mix); and security (street lighting, cameras, police patrols). Literature has revealed that pedestrians value factors directly affecting safety, such as road width, vehicle speed and volume, connectivity, and lighting conditions, higher than comfort-related factors. Pedestrian Safety Countermeasure Effects on Pedestrian Safety, Operations, and Quality of Service Literature has quantified the effects of only some countermeasures on pedestrian operations and LOS. This literature is mostly related to pedestrian signal timing strategies, such as LPIs (Fayish and Gross 2010, Kothuri et al. 2017, Saneinejad and Lo 2015) and exclusive pedestrian phases (Bechtel et al. 2004). Most other countermeasures implemented to improve pedestrian safety have not been studied for their effects on pedestrian operations. However, the effects of some of these countermeasures can be predicted based on how they change factors related to the HCM’s pedestrian LOS methods. A summary of the expected effects of the countermeasures listed above is shown in Table A15. Table A15. Effects of Pedestrian Safety Countermeasures on Pedestrian Operations and LOS at Intersections Factors Effect on Pedestrian Operations Roadway Segments Sidewalks Improves pedestrian LOS Signalized Intersections Red clearance interval Potentially decreases pedestrian LOS due to increased pedestrian delay at intersection, if the cycle length is increased as a result. Exclusive phasing, pedestrian scramble Reduced pedestrian LOS due to increased pedestrian delay at intersection (Bechtel et al. 2004). Increased pedestrian delays are caused by longer cycle lengths. Additionally, can increase pedestrian noncompliance due to higher delays (Bechtel et al. 2004). Leading pedestrian interval Reduces pedestrian LOS due to increased pedestrian delay at intersection. (Fayish and Gross 2010; Kothuri et al. 2017; Saneinejad and Lo 2015). Increased pedestrian delays could be caused by longer cycle lengths. Pedestrian push buttons Potentially improves pedestrian LOS due to decreased pedestrian delay at pedestrian-actuated signals. Pedestrian countdown timers No direct impact on pedestrian LOS. May reduce pedestrian delay if pedestrians determine they have sufficient time to cross during flashing DON’T WALK. Curb extensions No direct impact on pedestrian LOS. If the reduced crossing distance allows the cycle length to be reduced, could improve LOS. Parking removal near the intersection Potentially reduces LOS due to removal of barrier between moving traffic and pedestrians Curb ramps No impact on pedestrian LOS Median refuge If used to create a multi-stage pedestrian crossing, may increase pedestrian LOS due to added crossing delay.

216 Factors Effect on Pedestrian Operations Unsignalized Crossings Marked crosswalks May improve driver yielding rate, thereby reducing crossing delay. If a crosswalk provides a new legal crossing opportunity that did not exist before, could improve pedestrian link and segment LOS due to the midblock crossing opportunity Advance YIELD/STOP signs May improve driver yielding rate High-visibility crosswalks May improve driver yielding rate Pedestrian hybrid beacons May improve pedestrian LOS by reducing street-crossing delay, compared to waiting for a gap in traffic Rectangular rapid-flashing beacons May improve driver yielding rate Median refuge, raised median If sufficiently wide to store pedestrians, allows for two-stage crossings, with overall lower delay (improves LOS). Illumination May improve driver yielding rate Pedestrian overpasses and underpasses May introduce extra walking distance into the street crossing Raised pedestrian crossings May improve driver yielding rate Current HCM Methods Simplified Versions of HCM Methods With the publication of the HCM 6th Edition (TRB 2016) and the Planning and Preliminary Engineering Applications Guide to the HCM (PPEAG) (Dowling et al. 2016), there are now two avenues available for providing pedestrian LOS measures based on pedestrian operations and pedestrian QOS. The PPEAG is intended for higher-level analyses that incorporate fewer and less-precise data inputs, while the HCM is intended for more-detailed analysis accounting for a full range of factors that influence operations or QOS. However, this ability to distinguish between levels of effort is not taken advantage of at present for pedestrian analyses, as the PPEAG generally refers readers to the HCM method (perhaps suggesting additional default values not included in the HCM) or to a simplification already suggested in the HCM (e.g., using a simpler link analysis instead of a full segment analysis). NCHRP Project 17-87 could potentially develop models with different levels of detail for different applications, with one set intended for incorporation in the PPEAG and the other incorporated in the HCM. Signalized Intersections The HCM currently provides LOS based on traveler perception (pedestrian intersection LOS score), but not based on operations (delay, crosswalk corner area, crosswalk circulation area), although it provided LOS ranges for operational measures prior to the HCM 2010. Given that speed- or delay-based measures of LOS are presented in many other parts of the HCM (including for pedestrians at unsignalized crossings), presenting both operational and traveler perception LOS measures could be considered. Alternatively, a combined LOS, where the reported LOS is the worse of the operational and traveler perception results, could be used, similar to what is currently done for urban street segments. The current method for estimating pedestrian delay at signalized crossings does not accurately account for multi-stage crossings, pedestrian-actuated signals, and non-random pedestrian arrivals. The current method for estimating required crosswalk width to provide a desired LOS assumes a constant pedestrian speed, whereas research has found that these speeds are variable, depending on a variety of factors. The

217 HCM exhibit relating ranges of pedestrian space values for queuing areas at street corners uses values for moving pedestrians (e.g., along sidewalks) rather than for standing pedestrians, which appears to be an error. The pedestrian intersection LOS score is insensitive to the following required inputs: pedestrian delay, right-turn-on-red volume, and permitted left-turn volume Urban Street Segments and Facilities The HCM provides a combined LOS based on both traveler perception (pedestrian segment LOS score) and operations (pedestrian space), with the lower of the two LOS results being reported as the segment LOS. The link component of the segment LOS score is relatively insensitive to the following required inputs: roadway lane widths, traffic speed, landscape buffer width, and sidewalk width (except when installing a sidewalk where none existed previously). Although pedestrian segment LOS is mathematically highly sensitive to the roadway crossing difficulty factor, for practical purposes it is difficult to devise a crossing treatment that will produce much of an effect on the segment LOS score. The model was developed from ratings of pedestrian facilities along urban and suburban collector and arterial streets; it is unknown how pedestrian ratings might differ for local streets or in rural areas. Very large changes in pedestrian flow rates or sidewalk effective widths are needed to observe a change in pedestrian space LOS. Except in areas with very high levels of pedestrian activity, pedestrian space LOS will typically be A or B, and thus not needed to evaluate an overall segment LOS. Although platoon flow is more likely to be the typical case in situations where it makes sense to evaluate pedestrian space LOS, average flow is presented as the default situation. It is unknown whether the shy distances used to determine effective sidewalk width vary by pedestrian volume, time of day, and/or adjacent land use. Although the HCM identifies that pedestrians will spill out of the designated pedestrian circulation area at pedestrian space LOS values better than F, no guidance is provided on a specific LOS where this effect starts to occur. Uncontrolled Crossings The HCM provides an operational LOS based on delay for street crossings where traffic on the street being crossed is uncontrolled; no traveler perception measure is provided. No LOS measure is provided for crossings where traffic is controlled (e.g., by a STOP sign), in part because NCHRP Project 03-70 found no effect of driveways and controlled side-street intersections on urban street segment pedestrian LOS. The HCM’s method for estimating delay is sensitive to many of the types of pedestrian safety countermeasures that could be considered for an uncontrolled crossing, by shortening the crossing distance, improving driver yielding rates, or both. However, there appear to be problems with the HCM’s yielding model that produce unreasonable estimates of delay. In addition, the model has not been field-tested. Other Intersection Forms The HCM provides no pedestrian LOS measure for a variety of intersection forms, including all-way stops, roundabouts, interchange ramp terminals, and A.I.I. At best, guidance is provided on adapting the signalized intersection or uncontrolled intersection method, as appropriate, to the intersection form.

218 References Akçelik & Associates Pty. Ltd. 2001. An Investigation of Pedestrian Movement Characteristics at Mid-Block Signalized Crossings. Technical Report. Greythorn, Victoria, Australia. Al-Azzawi, M., and R. Raeside. 2007. “Modeling pedestrian walking speeds on sidewalks.” Journal of Urban Planning and Development, Vol. 133, No. 3, pp. 211–219. Ali, A.T., A. Flannery, and C. Cristei. 2010. “Cumulative Logistic Regression Model for Pedestrian Level of Service Rating.” Paper 10-3945. Presented at the 89th Annual Meeting of the Transportation Research Board, Washington, DC. Al-Masaeid, H.R., T.I. Al-Suleiman, and D.C. Nelson. 1993. “Pedestrian speed–flow relationship for central business district areas in developing countries.” Transportation Research Record: Journal of the Transportation Research Board, No. 1396, pp. 69–74. Alsaleh, R., T. Sayed, and M. Zaki. 2018. “Assessing the Effect of Pedestrians Use of Cell Phones on their Walking Behavior: A Study Based on Automated Video Analysis.” Transportation Research Record: Journal of the Transportation Research Board. DOI: 10.1177/0361198118780708 Anciaes, P.R., and P. Jones. 2018. “Estimating preferences for different types of pedestrian crossing facilities.” Transportation Research Part F, No. 52, pp. 222–237. Araujo, G.P., and M. Braga. 2008. “Methodology for the Qualitative Evaluation of Pedestrian Crossings at Road Junctions with Traffic Lights.” Transportation, Vol. 35, pp. 539–557. Arshad, A., N. Bahari, W. Hashim, and A. Halim. 2016. “Gender Differences in Pedestrian Perception and Satisfaction on the Walkability of Kuala Lumpur City Center.” MATEC Web of Conferences, Vol. 47, 03003. Bacquie, R., D. Egan, and L. Ing. 2001. “Pedestrian Refuge Island Safety Audit.” Presented at 2001 ITE Spring Conference and Exhibit, Monterey, CA. Banerjee, I., and D. Ragland. 2007. “Evaluation of Countermeasures: A Study on the Effect of Impactable Yield Signs Installed at Four Intersections in San Francisco.” Paper 07-2947. Presented at the 86th Annual Meeting of the Transportation Research Board, Washington, DC. Barton & Loguidice, DPC, and Sprinkle Consulting, Inc. 2017. Geneva Active Transportation Plan. Geneva, N.Y. http://cityofgenevany.com/wp-content/uploads/2016/10/DRAFT_Geneva-Active-Transportation-Plan.pdf (As of Sept. 19, 2018). Basile, O, L. Persia, and D. Usami. 2010. “A methodology to assess pedestrian crossing safety.” European Transport Research Review, Vol. 2., pp. 129–137. Bechtel, A., K. MacLeod, and D. Ragland. 2004. “Pedestrian Scramble Signal in Chinatown Neighborhood of Oakland California: An Evaluation.” Transportation Research Record: Journal of the Transportation Research Board, No. 1878, pp. 19–26. Bella, F., and M. Silvestri. 2015. “Effects of Safety Measures on Driver’s Speed Behavior at Pedestrian Crossings.” Accident Analysis & Prevention, Vol. 83, pp. 111–124. Bennett, M.K., H. Manal, and R. Van Houten. 2014. “A Comparison of Gateway In-Street Sign Treatment to Other Driver Prompts to Increase Yielding to Pedestrians at Crosswalks.” Paper 14-0222. Presented at the 93rd Annual Meeting of the Transportation Research Board, Washington, DC. Bennett, S., A. Felton, and R. Akçelik. 2001. “Pedestrian Movement Characteristics at Signalized Intersections.” Presented at the 23rd Conference of the Australian Institutes of Transportation Research, Melbourne, Victoria, Australia. Bowman, B.L., and R. Vecellio. 1994. “Effects of Urban and Suburban Median Types on Both Vehicular and Pedestrian Safety.” Transportation Research Record, No. 1445, pp. 169–179. Bradbury, K., J. Stevens, L. Boyle, and S. Rutherford. 2012. “To Go or Not To Go: Pedestrian Behavior at Intersections with Standard Pedestrian Call Buttons.” Transportation Research Record: Journal of the Transportation Research Board, No. 2299, pp. 174–179. Brunson, C., A. Getman, S. Hostetter, and R. Viola. 2017. “Left-Turn Pedestrian and Bicycle Crash Study.” Paper 17- 05455. Presented at the Transportation Research Board 96th Annual Meeting, Washington DC. Camden, A., R. Buliung, L. Rothman, C. Macarthur, and A. Howard. 2011. “The impact of pedestrian countdown signals on pedestrian–motor vehicle collisions: a quasi-experimental study.” Injury Prevention, DOI: 10.1136/injuryprev-2011-040173.

219 Campbell, B.J., C.V. Zegeer, H.H. Huang, and M. Cynecki. 2003. A Review of Pedestrian Safety Research in the United States and Abroad. Report FHWA-RD-03-042. Federal Highway Administration, Office of Safety, Washington, DC. Carey, N. Establishing Pedestrian Walking Speeds. 2005. Portland State University, Portland, OR. https://www.westernite.org/datacollectionfund/2005/psu_ped_summary.pdf (As of Sept. 19, 2018). Charlier Associates, Inc.; K. Krizek; and A. Forsyth. 2012. The Colorado Mile Markers: Recommendations for Measuring Active Transportation. Kaiser Permanente, Denver, CO. Chattaraj, U., A. Seyfried, and P. Chakroborty. 2009. “Comparison of pedestrian fundamental diagram across cultures.” Advances in Complex Systems, Vol. 12, No. 03, pp. 393–405. Chen, L., C. Chen, and R. Ewing. 2014. “The relative effectiveness of signal related pedestrian countermeasures at urban intersections—Lessons from a New York City case study.” Transport Policy, Vol. 32, pp. 69–78. Chen, L., C. Chen, R. Ewing, C.E. McKnight, R. Srinivasan, and M. Roe. 2013. “Safety countermeasures and crash reduction in New York City—Experience and lessons learned.” Accident Analysis & Prevention, Vol. 50, pp. 312– 322. Chimba, D., D. Emaasit, C.R. Cherry, and Z. Pannell. 2014. “Patterning Demographic and Socioeconomic Characteristics Affecting Pedestrian and Bicycle Crash Frequency.” Paper 14-0600. Presented at the Transportation Research Board 93rd Annual Meeting, Washington DC. City of Madison. 1999. Year 2 Field Evaluation of Experimental “In-Street” Yield to Pedestrian Signs. Traffic Engineering Division, City of Madison Department of Transportation, Madison, WI. Cleland, V.J., A. Timperio, and D. Crawford. 2008. “Are perceptions of the physical and social environment associated with mothers' walking for leisure and for transport? A longitudinal study.” Preventive Medicine, Vol. 47, No. 2, pp. 188–193. Coffin, A., and J. Morral. 1995. “Walking Speeds of Elderly Pedestrians at Crosswalks.” Transportation Research Record, No. 1487, pp. 63–67. Cohen, J., E.J. Dearnaley, and C.E.M. Hansel. 1955. “The risk taken in crossing a road.” Journal of the Operational Research Society, Vol. 6, No. 3, pp. 120–128. Crompton, D.H. 1979. “Pedestrian delay, annoyance and risk: preliminary results from a 2 years study.” Presented at the 7th Summer Annual Meeting. Proceedings of Seminar J, PTRC, Warwick, U.K., pp. 275–299. Dahlstedt, S. 1978. “Walking Speeds and Walking Habits of Elderly People.” Proceedings of the International Conference on Transport for the Elderly and Handicapped, Cambridge, U.K., pp. 243–249. Das, S., C.F. Manski, and M.D. Manuszak. 2005. “Walk or wait? An empirical analysis of street crossing decisions.” Journal of Applied Econometrics, Vol. 20, No. 4, pp. 529–548. Dewar, R. 2002. Human Factors in Traffic Safety. Lawyers and Judges Publishing Company, Tucson, AZ. Diogenes, M., R. Greene-Roesel, L. Arnold, and D. Ragland. 2007. “Pedestrian Counting Methods at Intersections: A Comparative Study.” Transportation Research Record: Journal of the Transportation Research Board, No. 2002, pp. 26–30. Diogenes, M., and L. Lindau. 2010. “Evaluation of Pedestrian Safety at Midblock Crossings, Porto Alegre, Brazil.” Transportation Research Record: Journal of the Transportation Research Board, No. 2193, pp. 37–43. DiPietro, C. M., and L.E. King. 1970. “Pedestrian Gap-acceptance.” Highway Research Record 308, pp. 80–91. Domarad, J., P. Grisak, and J. Bolgar. 2013. “Improving Crosswalk Safety: Rectangular Rapid-Flashing Beacon (RRFB) Trial in Calgary.” Presented at ITE District 7 Canada Annual Meeting, Calgary, Alberta. Dougald, L.E. 2016. “Effectiveness of a Rectangular Rapid-Flashing Beacon at a Midblock Crosswalk on a High- Speed Urban Collector.” Transportation Research Record: Journal of the Transportation Research Board, No. 2562, pp. 36–44. Dowling, R., D. Reinke, A. Flannery, P. Ryus, M. Vandehey, T. Petritsch, B. Landis, N. Rouphail, and J. Bonneson. 2008. NCHRP Report 616: Multimodal Level of Service for Urban Streets. Transportation Research Board of the National Academies, Washington, DC. Dowling, R., A. Flannery, P. Ryus, T. Petritsch, B. Landis, and N. Rouphail. 2010. NCHRP Web-only Document 158: Field Test Results of the Multimodal Level of Service Analysis for Urban Streets. Transportation Research Board of the National Academies, Washington, DC. Dowling, R., P. Ryus, B. Schroeder, M. Kyte, F.T. Creasey, N. Rouphail, A. Hajbabaie, and D. Rhoades. 2016. NCHRP Report 825: Planning and Preliminary Engineering Applications Guide to the Highway Capacity Manual. Transportation Research Board, Washington, DC.

220 Dutata, N., Q. Zhang, and M. Kroneberger. 2014. “Impact of Intersection Design on Pedestrians’ Choice to Cross on Red.” Transportation Research Record: Journal of the Transportation Research Board, No. 2464, pp. 93–99. Eccles, K., R. Tao, and B. Mangum. 2004. “Evaluation of Pedestrian Countdown Signals in Montgomery County, Maryland.” Transportation Research Record: Journal of the Transportation Research Board, No. 1878, pp. 36– 41. Ellis, R., R. Van Houten, and J.L. Kim. 2007. “In-roadway ‘Yield to Pedestrians’ Signs: Placement Distance and Motorist Yielding.” Transportation Research Record: Journal of the Transportation Research Board, No. 2002, pp. 84–89. Elvik, R., and T. Vaa (Eds.). 2004. The Handbook of Road Safety Measures. Elsevier, Oxford, U.K. Ewing, R., M.R. King, S. Raudenbush, and O.J. Clemente. 2005. “Turning Highways into Main Streets: Two Innovations in Planning Methodology.” Journal of the American Planning Association, Vol. 71, No. 3, pp. 269– 282. Ewing, R. 2000. “Asking Transit Users about Transit-Oriented Design.” Transportation Research Record: Journal of the Transportation Research Board, No. 1735, pp. 19–24. Ewing, R., and R. Cervero. 2001. “Travel and the Built Environment: A Synthesis.” Transportation Research Record: Journal of the Transportation Research Board, No. 1780, pp. 87–114. Ewing, R., S. Handy, R.C. Brownson, O. Clemente, and E. Winston. 2006. “Identifying and measuring urban design qualities related to walkability.” Journal of Physical Activity and Health, Vol. 3, No. S1, pp. S223–S240. Fartash, H., P. Alluri, A. Gan, M. Lavasani, and K. Haleem. 2016. “Performance of Rectangular Rapid Flashing Beacons at Midblock Pedestrian Crossings.” Paper 16-1938. Presented at the Transportation Research Board 95th Annual Meeting, Washington DC. Fayish, A., and F. Gross. 2010. “Safety Effectiveness of Leading Pedestrian Intervals Evaluated by a Before–After Study with Comparison Groups.” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, pp. 15–22. Federal Highway Administration. 2000. Manual on Uniform Traffic Control Devices. U.S. Department of Transportation, Washington, DC. Federal Highway Administration. 2004. Toolbox of Countermeasures and Their Potential Effectiveness to Make Intersections Safer. Issue Brief FHWA-SA-014. U.S. Department of Transportation, Washington, DC. Federal Highway Administration. 2009. Manual on Uniform Traffic Control Devices. U.S. Department of Transportation, Washington, DC. Federal Highway Administration. 2011. Pedestrian and Bicycle Data Collection. U.S. Department of Transportation, Washington, DC. Federal Highway Administration. 2013. Traffic Monitoring Guide. Publication FHWA PL-13-015. U.S. Department of Transportation, Washington, DC. Federal Highway Administration. 2016. Road User Behaviors at Pedestrian Hybrid Beacons. TechBrief FHWA- HRT-16-039, U.S. Department of Transportation, Washington, DC. Feldman, M., J.G. Manzi, and M.F. Mitman. 2010. “Empirical Bayesian Evaluation of Safety Effects of High-visibility School (Yellow) Crosswalks in San Francisco, California.” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, pp. 8–14. Finnis, K.K., and D. Walton. 2008. “Field observations to determine the influence of population size, location and individual factors on pedestrian walking speeds.” Ergonomics, Vol. 51, No. 6, pp. 827–842. Fitzpatrick, K., and E.S. Park. 2010. Safety Effectiveness of the HAWK Pedestrian Crossing Treatment. Report FHWA HRT-10-042, Federal Highway Administration, Washington, DC. Fitzpatrick, K., M. Brewer, and S. Turner. 2006. “Another Look at Pedestrian Walking Speed.” Transportation Research Record: Journal of the Transportation Research Board, No. 1982, pp. 21–29. Fitzpatrick, K., S.T. Chrysler, R. Van Houten, W.W. Hunter, and S. Turner. 2011. Evaluation of Pedestrian and Bicycle Engineering Countermeasures: Rectangular Rapid-Flashing Beacons, HAWKs, Sharrows, Crosswalk Markings, and the Development of an Evaluation Methods Report. Report FHWA-HRT-11-039. Federal Highway Administration, Washington, DC. Fitzpatrick, K., S. Chrysler, V. Iragavarapu, and E. Park. 2011. “Detection Distances to Crosswalk Markings: Transverse Lines, Continental Markings, and Bar Pairs.” Transportation Research Record: Journal of the Transportation Research Board, No. 2250, pp. 1–10.

221 Fitzpatrick, K., S. Turner, M. Brewer, P. Carlson, N. Lalani, B. Ullman, N. Trout, E.S. Park, D. Lord, and J. Whitacre. 2006. TCRP Report 112/NCHRP Report 562: Improving Pedestrian Safety at Unsignalized Crossings. Transportation Research Board of the National Academies, Washington, DC. Fitzpatrick, K., V. Iragavarapu, M.A. Brewer, D. Lord, J. Hudson, R. Avelar, and J. Robertson. 2014. Characteristics of Texas Pedestrian Crashes and Evaluation of Driver Yielding at Pedestrian Treatments. Report FHWA/TX- 13/0-6702-1. Texas Department of Transportation, Austin. Fitzpatrick, K., M.A. Brewer, R. Avelar, and T. Lindheimer. 2017. “Will You Stop for Me? An Exploration of Characteristics Associated with a Driver’s Decision to Stop for a Pedestrian in a Crosswalk with a Rectangular Rapid-Flashing Beacon.” ITE Journal, Vol. 87, No. 3, pp. 36–41. Fitzpatrick, K., M.A: Brewer, and R. Avelar. 2014. “Driver Yielding at Traffic Control Signals, Pedestrian Hybrid Beacons, and Rectangular Rapid-flashing Beacons in Texas.” Transportation Research Record: Journal of the Transportation Research Board, No.2463, pp. 46–54. Foster, N., C. Monsere, and K. Carlos. 2014. “Evaluating Driver and Pedestrian Behaviors at Enhanced, Multilane, Midblock Pedestrian Crossings: Case Study in Portland, Oregon.” Transportation Research Record: Journal of the Transportation Research Board, No. 2464, pp. 59–66. Fruin, J. J. 1971a. “Designing for Pedestrians: A Level of Service Concept.” Highway Research Record, No. 355, pp. 1–15. Fruin, J.J. 1971b. Pedestrian Planning and Design. Metropolitan Association of Urban Designers and Environmental Planners, New York. Fruin, J.J., and G.P. Benz. 1984. “Pedestrian Time–Space Concept for Analyzing Corners and Crosswalks.” Transportation Research Record, No. 959, pp. 18–24. Gan, A., J. Shen, and A. Rodriguez. 2005. Update of Florida Crash Reduction Factors and Countermeasures to Improve the Development of District Safety Improvement Projects. Florida DOT, Tallahassee. Gårder, P.E. 2004. “The Impact of Speed and Other Variables on Pedestrian Safety in Maine.” Accident Analysis & Prevention, Vol. 36, No. 4, pp. 533–542. Gates, T. J., D.A. Noyce, A.R. Bill, N. Van Ee, and T.J. Gates. 2006. “Recommended Walking Speeds for Pedestrian Clearance Timing Based on Pedestrian Characteristics.” Paper 06-1826. Presented at the 87th Annual Meeting of the Transportation Research Board, Washington, DC, 2006. Gates, T.J., P.T. Savolainen, S. Stapleton, T. Kirsch, and S. Miraskar. 2016. Development of Safety Performance Functions and Other Decision Support Tools to Assess Pedestrian and Bicycle Safety. Report TRCLC 14-6. Transportation Research Center for Livable Communities, Wayne State University, Detroit, MI. Gedafa, D.S., B, Kaemingk, B. Mager, J. Pape, M. Tupa, and T. Bohan. 2014. “Impacts of Alternative Yield Sign Placement on Pedestrian Safety.” Transportation Research Record: Journal of the Transportation Research Board, No. 2464, pp. 11–19. Gerilla, G.P. 1995. “Proposed level of service standards for walkways in Metro Manila.” Journal of East Asia Society of Transportation Studies, Vol. 1, No. 3, pp. 1041–1060. Godfrey, D., and T. Mazella. 1999. “Kirkland's Experience with In-Pavement Flashing Lights at Crosswalks.” Presented at the ITE/IMSA Annual Meeting, Lynnwood, WA. Greene-Roesel, R., M.C. Diogenes, D.R. Ragland, and L.A. Lindau. 2008. Effectiveness of a Commercially Available Automated Pedestrian Counting Device in Urban Environments: Comparison with Manual Counts. Safe Transportation Research & Education Center, University of California, Berkeley. Griffin, G., K. Nordback, T. Götschi, E. Stolz, and S. Kothuri. 2014. Transportation Research E-Circular E-C183: Monitoring Bicyclist and Pedestrian Travel and Behavior: Current Research and Practice. Transportation Research Board of the National Academies, Washington, DC. Gross, F., B. Persaud, and C. Lyon. 2010. Guide for Developing Quality Crash Modification Factors. Report FHWA- SA-10-032. Federal Highway Administration, Washington, DC. Guerrier, J.H., and S.C. Jolibois. 1998. “The Safety of Elderly Pedestrians at Five Urban Intersections in Miami.” Presented at the 42nd Annual Meeting of the Human Factors and Ergonomics Society, Santa Monica, CA. Gupta, A., and N. Pundir. 2015. “Pedestrian flow characteristics studies: A review.” Transport Reviews, Vol. 35, No. 4, pp. 445–465. Hakkert, A.S., V. Gitelman, and E. Ben-Shabat. 2002. “An evaluation of crosswalk warning systems: effects on pedestrian and vehicle behaviour.” Transportation Research Part F: Traffic Psychology and Behaviour, Vol. 5, No. 4, pp. 275–292.

222 Hankin, B.D., and R.A. Wright. 1958. “Passenger Flows in Subways.” OR: Journal of the Operational Research Society, No. 9, pp. 81–88. Harkey, D.L., R. Srinivasan, J. Baek, B. Persaud, C. Lyon, F.M. Council, K. Eccles, N. Lefler, F. Gross, E. Hauer, and J. Bonneson. 2008. NCHRP Research Results Digest 299: Crash Reduction Factors for Traffic Engineering and Intelligent Transportation System (ITS) Improvements: State-of-Knowledge Report. Transportation Research Board of the National Academies, Washington, DC. Harwood, D.W., K.M. Bauer, K.R. Richard, D.K. Gilmore, J.L. Graham, I.B. Potts, D.J. Torbic, and E. Hauer. 2008. NCHRP Web-Only Document 129: Pedestrian Safety Prediction Methodology. Transportation Research Board of the National Academies, Washington, DC. Helbing, D., A. Johansson, and H.Z. Al-Abideen. 2007. “Dynamics of Crowd Disasters: An Empirical Study.” Physical Review E, Statistical, Nonlinear, and Soft Matter Physics, Vol. 75, No. 4, article ID 046109. Hine, J., and J. Russell. 1996. “The impact of traffic on pedestrian behaviour 2: Assessing the traffic barrier on radial routes.” Traffic Engineering and Control, Vol. 37, No. 2, pp. 81–85. Hipp, J., D. Adlakha, A. Eyler, B. Chang, and R. Pless. 2013. “Emerging Technologies: Webcams and Crowd- Sourcing to Identify Active Transportation.” American Journal of Preventive Medicine, Vol. 44, No. 1, pp. 96– 97. Hong, J., and D. Park. 2017. “The Impact of Trip Purposes and Built Environment on Levels of Pedestrian Satisfaction: Applications of Mixed Process Ordered Probit and Two-Stage Residual Inclusion Estimations.” Paper 17-02956. Presented at the 97th Annual Meeting of the Transportation Research Board, Washington, DC. Hua, J., N. Gutierrez, I. Banerjee, F. Markowitz, and D.R. Ragland. 2009. “San Francisco PedSafe II Project Outcomes and Lessons Learned. Paper 09-2988. Presented at the 88th Annual Meeting of the Transportation Research Board, Washington, DC. Huang, H. 2000. An Evaluation of Flashing Crosswalks in Gainesville and Lakeland. Florida Department of Transportation Tallahassee. Huang, H.F., and M.J. Cynecki. 2001. The Effects of Traffic Calming Measures on Pedestrian and Motorist Behavior. Report FHWA-RD-00-104. Federal Highway Administration, Washington, DC. Hudson, J., T. Qu, and S. Turner. 2010. Forecasting Bicycle and Pedestrian Usage and Research Data Collection Equipment. Report P2009330. Texas Transportation Institute, College Station. Huff, H., and R. Liggett. 2014. “The Highway Capacity Manual’s Method for Calculating Bicycle and Pedestrian Levels of Service: The Ultimate White Paper.” University of California, Los Angeles. Hughes, R., H. Huang, C.V. Zegeer, and M.J. Cynecki. 2001. Evaluation of Automated Pedestrian Detection at Signalized Intersections. Report FHWA-RD-00-097. Federal Highway Administration, Washington, DC. Hung, W.T., A. Manadhar, and S.A. Ranasinghege. 2010. “A Walkability Survey in Hong Kong.” Presented at the 12th International Conference on Mobility and Transport for the Elderly, Hong Kong. Hunter, W.W., R. Srinivasan, and C.A. Martell. 2012. “Evaluation of the Rectangular Rapid Flash Beacon at a Pinellas Trail Crossing in St. Petersburg, Florida.” Transportation Research Record: Journal of the Transportation Research Board, No. 2314, pp. 7–13. Injury Surveillance Workgroup 8 (ISW8). 2017. Consensus Recommendations for Pedestrian Injury Surveillance. Safe States Alliance, Atlanta, GA. Ishaque. M., and R. Noland. 2008. “Behavioral Issues in Pedestrian Speed Choice and Street Crossing Behavior. A Review.” Transport Reviews, Vol. 28, No. 1, pp. 61–85. Ivan, J.N., K. McKernan, Y. Zhang, N. Ravishanker, and S.A. Mamun. “A study of pedestrian compliance with traffic signals for exclusive and concurrent phasing.” Accident Analysis & Prevention, Vol. 98, pp. 157–166. Jackson, K.N., S.W. O’Brien, S.E. Searcy, and S.E. Warchol. 2017. Quality Assurance and Quality Control Processes for a Large-Scale Bicycle and Pedestrian Volume Data Program. Transportation Research Record: Journal of the Transportation Research Board, No. 2644, pp. 19–29. Jang, K., S. Park, S. Kang, K. Song, S. Kang, and S. Chung. 2013 “Evaluation of Pedestrian Safety: Pedestrian Crash Hot Spots and Risk Factors for Injury Severity.” Transportation Research Record: Journal of the Transportation Research Board, No. 2393, pp. 104–116. Jensen, S.U. 2013. “Pedestrian and Bicycle Level of Service at Intersections, Roundabouts and Other Crossings.” Paper 13-0950. Presented at the 93rd Annual Meeting of the Transportation Research Board, Washington DC. Jia, H., L. Yang, and T. Tang. 2009. “Pedestrian flow characteristics analysis and model parameter calibration in comprehensive transport terminal.” Journal of Transportation Systems Engineering and Information Technology, Vol. 9, No. 5, pp. 117–123.

223 Jung, H., S. Lee, H. Kim, and J. Lee. 2017. “Does Improving the Physical Street Environment Create Satisfactory and Active Streets? Evidence from Seoul’s Design Street Project.” Transportation Research Part D, Vol 50, pp. 269– 279. Kamyab, A., S. Andrle, D. Kroeger, and D. Heyer. 2003. “Methods to Reduce Traffic Speed in High-Pedestrian Rural Areas.” Transportation Research Record: Journal of the Transportation Research Board, No. 1828, pp. 31–37. Karkee, G.J., S.S. Nambisan, S.S. Pulugurtha, and A. Singh. 2006. “An Evaluation of the Effectiveness of an In- pavement Flashing Light System.” Paper 06-1972. Presented at the 85th Annual Meeting of the Transportation Research Board, Washington, DC. Kattan, L., S. Acharjee, and R. Tay. 2009. “Pedestrian Scramble Operations: Pilot Study in Calgary, Alberta, Canada.” Transportation Research Record: Journal of the Transportation Research Board, No. 2140, pp. 79–84. Kim, S., S. Park, and J. Lee. 2014. “Meso- or Micro-Scale? Environmental Factors Influencing Pedestrian Satisfaction.” Transportation Research Part D, Vol. 30, pp. 10–20. King, M.R. 1999. “Calming New York City Intersections.” Transportation Research Board Circular, E-CO19: Urban Street Symposium, Issue Number 501, p. 15. King, M.R., J.A. Carnegie, and R. Ewing. 2003. “Pedestrian Safety Through a Raised Median and Redesigned Intersections.” Transportation Research Record: Journal of the Transportation Research Board, No. 1828, pp. 56–66. Kittelson & Associates, Inc.; Center for Urban Transportation and Research; AECOM, Inc.; and Write Rhetoric. 2018. Multimodal Corridor Access Management Analysis Guide. Final Draft, NCHRP Project 03-120. Portland, OR. Kittelson & Associates, Inc.; Institute for Transportation Research and Education; Toole Design Group; Accessible Design for the Blind; and ATS Americas. 2017. NCHRP Project 07-25: Guide for Pedestrian and Bicycle Safety at Alternative Intersections and Interchanges. Interim Report. Wilmington, NC. Kittelson & Associates, Inc. 2015. “Proposed Person Delay Methodology.” Technical Memorandum. Portland Bureau of Transportation, Portland, OR. Knoblauch, R.L., M. Pietrucha, and M. Nitzburg. 1996. “Field Studies of Pedestrian Walking Speed and Start-Up Time.” Transportation Research Record, No. 1538, pp. 27–38. Kothuri, S., K. Clifton, and C. Monsere. 2014. “Insights into Pedestrian Attitudes and Perceptions of Delay, Safety and Crossing Decisions at Signalized Intersections.” Paper 14-4229. Presented at the 93rd Annual Meeting of the Transportation Research Board, Washington, DC. Kothuri, S., T. Reynolds, C. Monsere, and P. Koonce. 2012. “Preliminary Development of Methods to Automatically Gather Bicycle Counts and Pedestrian Delay at Signalized Intersections.” Paper 12-2107. Presented at the Transportation Research Board Annual Meeting, Washington, D.C. Kothuri, S.M., A. Kading, E.J. Smaglik, and C. Sobie. 2017. Improving Walkability Through Control Strategies at Signalized Intersections. Report NITC-RR-782. Transportation Research and Education Center, Portland, OR. Kotkar, K.L., R. Rastogi, and S. Chandra. 2010. “Pedestrian Flow Characteristics in Mixed Flow Conditions.” Journal of Urban Planning and Development, Vol. 136, No. 3, pp. 23–33. Lagerwey, P.A., M.J. Hintze, J.B. Elliott, J.L. Toole, R.J. Schneider, and Kittelson & Associates, Inc. 2015. NCHRP Report 803: Pedestrian and Bicycle Transportation along Existing Roads—ActiveTrans Priority Tool Guidebook. Transportation Research Board of the National Academies, Washington, D.C, Lam, W.H.K., J.F. Morrall, and H. Ho. 1995. “Pedestrian Flow Characteristics in Hong Kong.” Transportation Research Record: Journal of the Transportation Research Board, No. 1487, pp. 56–62. Landis, B.W., V.R. Vattikuti, R.M. Ottenberg, D.S. McLeod, and M. Guttenplan. 2001. “Modeling the Roadside Walking Environment: Pedestrian Level of Service.” Transportation Research Record: Journal of the Transportation Research Board, No. 1773, pp. 82–88. LaPlante, J.N., and T.P. Kaeser. 2004. “The Continuing Evolution of Pedestrian Walking Speed Assumptions.” ITE Journal, Vol. 74, No. 9, pp. 32-40. Lee, C., and M. Abdel-Aty. 2005. “Comprehensive analysis of vehicle–pedestrian crashes at intersections in Florida.” Accident Analysis & Prevention, Vol. 37, No. 4, 2005, pp. 775–786. Lee, Y. C. 2005. Pedestrian walking and choice behavior on stairways and escalators in public transport facilities. Master of science thesis. University of Technology, Delft, The Netherlands. Levasseur, M., and G. Brisbane. 2011. Trial of Pedestrian Countdown Timers in the Sydney CBD. Final Report. Roads and Traffic Authority of New South Wales, Australia. Lindsey, G., S. Hankey, X. Wang, and J. Chen. 2013. The Minnesota Bicycle and Pedestrian Counting Initiative: Methodologies for Non-motorized Traffic Monitoring. Minnesota Department of Transportation, St. Paul.

224 Loukaitou-Sideris, A., R. Liggett, and H.G. Sung. 2007. “Death on the Crosswalk: A Study of Pedestrian–Automobile Collisions in Los Angeles.” Journal of Planning Education and Research, Vol. 26, No. 3, pp. 338–351. Lovejoy, K., F. Markowitz, and J. Montufar. “Use of Video for Automated Pedestrian Detection and Signal-Timing Extension: Results from a Pilot Installation in San Francisco, California.” Paper 12-4638, Presented at the 91st Annual Meeting of the Transportation Research Board, Washington, DC. Macdonald, E., N. Szibbo, W. Eisenstein, and L. Mozingo. 2018. “Quality-of-service: Toward a standardized rating tool for pedestrian quality of urban streets.” Journal of Urban Design, Vol. 23, pp. 71–93. Malenfant, L., and R. Van Houten. 1989. “Increasing the Percentage of Drivers Yielding to Pedestrians in Three Canadian Cities with a Multifaceted Safety Program.” Health Education Research, Vol. 5, No. 2, pp. 275–279. Markowitz, F., S. Sciortino, J.L. Fleck, and B.M. Yee. 2006. “Pedestrian Countdown Signals: Experience with an Extensive Pilot Installation.” ITE Journal, Vol. 76, No. 1, pp. 43-48. Martin, A. Factors Influencing Pedestrian Safety: A Literature Review. 2006. Report PPR241. TRL, Ltd., Berkshire, U.K. Martinez, L., and A. Barros. 2014. “Understanding the Factors that Influence Pedestrian Environment Quality.” Paper 14-1636. Presented at the 93rd Annual Meeting of the Transportation Research Board, Washington, DC. May, A.D. 1990. Traffic Flow Fundamentals. Prentice Hall, Englewood Cliffs, NJ, pp. 315–317. McMahon P., C. Duncan, J. Stewart, C. Zegeer, and A. Khattak. 1999. “Analysis of Factors Contributing to ‘Walking Along Roadway’ Crashes.” Transportation Research Record: Journal of the Transportation Research Board, No. 1674, pp. 41-48. Mead, J., C. Zegeer, and M. Bushell. 2013. Evaluation of Pedestrian-Related Roadway Measures: A Summary of Available Research. Report DTFH61-11-H-00024. Federal Highway Administration, Washington, DC. Mekuria, M.C., P.G. Furth, and H. Nixon. 2012. Low-Stress Bicycling and Network Connectivity. Report 11-19. Mineta Transportation Institute, San Jose, CA. Milazzo, J.S., N.M. Rouphail, J.E. Hummer, and D.P Allen. 1999. “Quality of Service for Interrupted-Flow Pedestrian Facilities in Highway Capacity Manual 2000.” Transportation Research Record, No. 1678, pp. 25–31. Miranda-Moreno, L.F., P. Morency, and A.M. El-Geneidy. 2011. “The link between built environment, pedestrian activity and pedestrian–vehicle collision occurrence at signalized intersections.” Accident Analysis & Prevention, Vol. 43, No. 5, pp. 1624–1634. Mishra, S. 2015. “Rectangular Rapid Flashing Beacons (RRFB) Pilot Project.” Presented at the 2015 Conference and Exhibition of the Transportation Association of Canada, Charlottetown, P.E.I., Canada. Mitman, M.F., D.R. Ragland, and C.V. Zegeer. 2008. “The Marked Crosswalk Dilemma: Uncovering Some Missing Links in a 35-Year Debate.” Transportation Research Record: Journal of the Transportation Research Board, No. 2073, pp. 86–93. Monsere, C., M. Figliozzi, S. Kothuri, A. Razmpa, and D. Hazel. 2017. Safety Effectiveness of Pedestrian Crossing Enhancements. Final Report, SPR 778. Oregon Department of Transportation, Salem. Monsere, C., Y. Wang, H. Wang, and C. Chen. 2017. Risk Factors for Pedestrian and Bicycle Crashes. Final Report, SPR 779. Oregon Department of Transportation, Salem. Montufar, M., J. Arango, M. Porter, and S. Nakagawa. 2007. “Pedestrians’ Normal Walking Speed and Speed When Crossing a Street.” Transportation Research Record: Journal of the Transportation Research Board, No. 2002, pp. 90–97. Moore, R.L. 1953. “Pedestrian choice and judgement.” OR: Journal of the Operational Research Society, Vol. 4, pp. 3–10. Morrall, J.F., L.L. Ratnayake, and P.N. Seneviratne. 1991. “Comparison of Central Business District Pedestrian Characteristics in Canada and Sri Lanka.” Transportation Research Record, No. 1294, pp. 57–61. Municipal Planning Association. 1923. Transit: A Part of the Pittsburgh Plan. Report No. 3. Pittsburgh, PA. Nabavi Niaki, M.S., T. Fu. N. Saunier, L.F. Miranda-Moreno, L. Amador, and J. Bruneau. 2016. “Road Lighting Effects on Bicycle and Pedestrian Accident Frequency: Case Study in Montreal, Quebec, Canada.” Transportation Research Record: Journal of the Transportation Research Board, No. 2555, pp. 86–94. Nambisan, S., S. Pulugurtha, V. Vasudevan, M. Dangeti, and V. Virupaksha. 2009. “Effectiveness of Automatic Pedestrian Detection Device and Smart Lighting on Pedestrian Safety.” Transportation Research Record: Journal of the Transportation Research Board, No. 2140, pp. 27–34. Navin, F.P.D., and Wheeler, R.J. 1969. “Pedestrian Flow Characteristics.” Traffic Engineering, Vol. 39, No. 4, pp. 30–36.

225 Nazir, M.I., S.K. Adhikary, Q.S. Hossain, and S.A. Ali. 2012. “Pedestrian flow characteristics in Khulna metropolitan city, Bangladesh.” Journal of Engineering Science, Vol. 3, No. 1, pp. 25–31. Nitzburg, M., and R. Knoblauch. 2001. An Evaluation of High-Visibility Crosswalk Treatments—Clearwater, Florida. Report FHWA-RD-00-105, Federal Highway Administration, Washington, DC. Nordback, K., S. Kothuri, T. Petritsch, P. McLeod, E. Rose, and H. Twaddell. 2016. Exploring Pedestrian Counting Procedures: A Review and Compilation of Existing Procedures, Good Practices, and Recommendations. Report FHWA-HPL-16-026. Federal Highway Administration, Washington, DC. Nordback, K., M. Sellinger, and T. Phillips. 2017. Estimating Walking and Bicycling at the State Level. Report NITC- RR-708. National Institute for Transportation and Communities, Portland, OR. New York City Department of City Planning. 2006. Pedestrian Level of Service Study: Phase I. New York. Obeid, H., H. Abkarian, M. Abou-Zeid, and I. Kaysi. 2017. “Analyzing driver-pedestrian interaction in a mixed-street environment using a driving simulator.” Accident Analysis & Prevention, Vol. 108, pp. 56–65. Oeding, D. 1963. “Verkehrsbelastung und Dimensionierung von Gehwegen und anderen Anlagen des Fußgängerverkehrs” (“Traffic Flow and Dimensioning of Sidewalks and other Pedestrian Facilities”). Strassenbau und Strassenterkehrstechnik, No. 22. Federal Transportation Ministry, Road Construction Dept., Bonn, West Germany. Older, S. J. 1968. “Movement of pedestrians on footways in shopping streets.” Traffic Engineering and Control, Vol. 10, pp. 160–163. Oregon Department of Transportation. 2017. Analysis Procedures Manual. Version 2, updated December 2017. Transportation Planning Analysis Unit, Salem. Palamara, P., and M. Broughton. 2013. “An Investigation of Pedestrian Crashes at Traffic Intersections in the Perth Central Business District.” Journal of Public Health, Vol. 93, pp. 1456–1463. Park, H.J., Y. Weili, Y. Weizhe, W. Ilana, and S. Ahmed. 2014. “Investigation of Pedestrian Crossing Speeds at Signalized Intersections with Heavy Pedestrian Volumes.” Transportation Research Record: Journal of the Transportation Research Board, No. 2463, pp. 62–69. Parida, P., and M. Parida. 2011. “Appreciation of Gender Differences in Development of Qualitative Level of Service for Sidewalks.” Presented at the 4th International Conference of Women’s Issues in Transportation, Irvine, CA. Pecchini, D, and F. Giuliani. 2015. “Street-Crossing Behavior of People with Disabilities.” Journal of Transportation Engineering, Vol. 141, No. 10, DOI:10.1061/(ASCE)TE.1943-5436.0000782. Pécheux, K., J. Bauer, and P. McLeod. 2009. Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reducing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures: Final System Impact Report. Federal Highway Administration, Washington, DC. Peters, D., L. Kim, R. Zaman, G. Haas, J. Cheng, and S. Ahmed. 2015. “Pedestrian Crossing Behavior at Signalized Intersections in New York City.” Transportation Research Record: Journal of the Transportation Research Board, No. 2519, pp. 179–188. Petritsch, T.A., B.W. Landis, P.S. McLeod, H.F. Huang, S. Challa, and M. Guttenplan. 2005. “Level-of-Service Model for Signalized Intersections.” Transportation Research Record: Journal of the Transportation Research Board, No. 1939, pp. 55–62. Poch, M., and F. Mannering. 1996. “Negative Binomial Analysis of Intersection-Accident Frequencies.” Journal of Transportation Engineering, Vol. 122 No. 2, pp. 105–113. Polus, A., J.L. Schofer, and A. Ushpiz. 1983. “Pedestrian Flow and Level of Service.” Journal of Transportation Engineering, Vol. 109, pp. 46–56. Pulugurtha, S.S., and A.P. Nujjetty. 2011. “Crash Estimation Models for Intersections.” Proceedings, Transportation and Development Institute Congress, pp. 1306–1314. Pulugurtha, S.S., and D.R. Self. 2013. “Pedestrian and Motorists’ Actions at Pedestrian Hybrid Beacon Sites: Findings from a Pilot Study.” International Journal of Injury Control and Safety Promotion, pp. 1–10. Pulugurtha, S.S., and V.R. Sambhara. 2011. “Pedestrian crash estimation models for signalized intersections.” Accident Analysis & Prevention, Vol. 43, No. 1, pp. 439–446. Pulugurtha, S.S., V. Vasudevan, S.S. Nambisan, and M.R. Dangeti. 2012. “Evaluating the Effectiveness of Infrastructure-Based Countermeasures on Pedestrian Safety.” Transportation Research Record: Journal of the Transportation Research Board, No. 2299, pp. 100–109. Pushkarev, B., and J. Zupan. 1975. Urban Space for Pedestrians. MIT Press, Cambridge, MA.

226 Raad, N., and M.I. Burke. 2018. “What Are the Most Important Factors for Pedestrian Level of Service Estimation? A Systematic Review of the Literature.” Transportation Research Record: Journal of the Transportation Research Board, DOI: 10.1177/0361198118790623. Rajendran, B., M. Parida, and S. Sapnani. 2018. “Modeling Perceived Pedestrian Level of Service of Sidewalks: An Application of Structural Equation Modeling.” Paper 18-00815. Presented at the 97th Annual Meeting of the Transportation Research Board, Washington DC. Rastogi, R., T. Ilango, and S. Chandra. 2013. “Pedestrian flow characteristics for different pedestrian facilities and situations.” European Transport, Vol. 53, Paper No. 6. Reddy, V., T. Datta, D. McAvoy, P. Savolainen, M. Abdel-Aty, and S. Pinapaka. 2008. Evaluation of Innovative Safety Treatments. Florida Department of Transportation, Tallahassee. Retting, R.A., J.F. Chapline, and A.F. Williams. 2002. “Changes in crash risk following re-timing of traffic signal change intervals.” Accident Analysis & Prevention, Vol. 34, No. 2, pp. 215–220. Ross, J., D. Serpico, and R. Lewis. 2011. Assessment of Driver Yielding Rates Pre- and Post-RRFB Installation, Bend, Oregon. Report FHWA-OR-RD 12-05. Oregon Department of Transportation, Salem. Rouphail, N., J. Hummer, J. Milazzo, and D. Allen. 1998. Capacity Analysis of Pedestrian and Bicycle Facilities: Recommended Procedures for the “Pedestrians” Chapter of the Highway Capacity Manual. Report FHWA-RD- 98-107. Federal Highway Administration, Washington, DC. Russo, B.J., E. James, C.Y. Aguilar, and E.J. Smaglik. 2018. “Pedestrian Behavior at Signalized Intersection Crosswalks: Observational Study of Factors Associated with Distracted Walking, Pedestrian Violations, and Walking Speed.” Transportation Research Record: Journal of the Transportation Research Board, DOI: 10.1177/0361198118759949. Ryan, S., and G. Lindsey. 2013. Counting Bicyclists and Pedestrians to Inform Transportation Planning. Robert Wood Johnson Foundation, Princeton, NJ. Ryus, P., E. Ferguson, K.M. Laustsen, R.J. Schneider, F.R. Proulx, T. Hull, and L. Miranda-Moreno. 2014. NCHRP Report 797: Guidebook on Pedestrian and Bicycle Volume Data Collection. Transportation Research Board of the National Academies, Washington, DC. Ryus, P., F.R. Proulx, R.J. Schneider, T. Hull, and L.F. Miranda-Moreno. 2014a. NCHRP Web Document 205: Methods and Technologies for Pedestrian and Bicycle Volume Data Collection. Transportation Research Board of National Academies, Washington, DC. Ryus, P., A. Butsick, F.R. Proulx, R.J. Schneider, and T. Hull. 2016. NCHRP Web Document 229: Methods and Technologies for Pedestrian and Bicycle Volume Data Collection: Phase 2. Transportation Research Board of the National Academies, Washington, DC. Sahani, R., V. Devi, and P. Bhuyan. 2017. “Defining Pedestrian Level of Service at Signalized Intersections through Modeled Parameters and SOM Clustering.” International Journal for Traffic and Transport Engineering, Vol. 7, No. 4, pp. 534–548. Said, M., M. Abou-Zeid, and I. Kaysi. 2016. “Modeling Satisfaction with the Walking Environment: The Case of an Urban University Neighborhood in a Developing Country.” Journal of Urban Planning and Development, DOI: 10.1061/(ASCE)UP.1943-5444.0000347. Saneinejad, S., and J. Lo. 2015. “Leading Pedestrian Interval: Assessment and Implementation Guidelines.” Transportation Research Record: Journal of the Transportation Research Board, No. 2519, pp. 85–94. Sarkar, A.K., and Janardhan, K.S.V.S. 2001. “Pedestrian flow characteristics at an intermodal transfer terminal in Calcutta.” World Transport Policy and Practice, Vol. 7, pp. 32–38. Schneider, R.J., A.J. Khattak, and C.V. Zegeer. 2001. “Method of Improving Pedestrian Safety Proactively with Geographic Information Systems: Example From a College Campus.” Transportation Research Record, No.1773, pp. 97–107. Schneider, R., R.M. Ryznar, and A.J. Khattak. 2004. “An accident waiting to happen: A spatial approach to proactive pedestrian planning. Accident Analysis & Prevention, Vol. 36 No. 2, pp. 193–211. Schneider, R., R. Patton, J. Toole, and C. Raborn. 2005. Pedestrian and Bicycle Data Collection in United States Communities: Quantifying Use, Surveying Users, and Documenting Facility Extent. Federal Highway Administration, Washington, DC. Schneider, R., L. Arnold, and D. Ragland. 2009. “Methodology for Counting Pedestrians at Intersections: Use of Automated Counters to Extrapolate Weekly Volumes from Short Manual Counts.” Transportation Research Record: Journal of the Transportation Research Board, Vol. 2140, pp. 1–12.

227 Schneider, R., M. Diogenes, L. Arnold, V. Attaset, J. Griswold, and D. Ragland. 2010. “Association Between Roadway Intersection Characteristics and Pedestrian Crash Risk in Alameda County, California.” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, pp. 41–51. Schrock, S.D., and B. Bundy. 2008. “Pedestrian Countdown Timers: Do Drivers Use Them to Increase Safety or Increase Risk Taking?” Paper 08-2203. Presented at the 87th Annual Meeting of the Transportation Research Board, Washington, DC. Senserrick, T., S. Boufous, L. De Rome, R. Ivers, and M. Stevenson. 2014. “Detailed Analysis of Pedestrian Casualty Collisions in Victoria, Australia.” Traffic Injury Prevention, Vol. 15, Sup. 1, pp. S197–S205. Seyfried, A., O. Passon, B. Steffen, M. Boltes, T. Rupprecht, and W. Klingsch. 2009. “New insights into pedestrian flow through bottlenecks.” Transportation Science, Vol. 43, pp. 395–406. Sharifi, M.S., D. Stuart, K. Christensen, and A. Chen. 2015. “Traffic Flow Characteristics of Heterogeneous Pedestrian Stream Involving Individuals with Disabilities.” Transportation Research Record: Journal of the Transportation Research Board, No. 2537, pp. 111–125. Sharma, A., J. Schmitz, A. Khattak, and V. Singh. 2012. “The Impact of Microscopic Factors in Evaluating the Effects of Pedestrian Countdown Timers on Pedestrian Decisions.” Paper 12-3029. Presented at the 91st Annual Meeting of the Transportation Research Board, Washington, DC. Shawky, M.A., A.M. Garib, and H. Al-Harthei. 2014. “The Impact of Road and Site Characteristics on the Crash-Injury Severity of Pedestrian Crashes.” Advances in Transportation Studies, 2014 Special Issue, Vol. 1, pp. 27–36. Shurbutt, J., and R. Van Houten. 2010. Effects of Yellow Rectangular Rapid-Flashing Beacons on Yielding at Multilane Uncontrolled Crosswalks. Report FHWA-HRT-10-043. Federal Highway Administration, Washington, DC. Staplin, L., K. Lococo, S. Byington, and D. Harkey. 2001. Guidelines and Recommendations to Accommodate Older Drivers and Pedestrians. Publication FHWA-RD-01-051. Federal Highway Administration, McLean, VA. Strong, C., and M. Kumar. 2006. Safety Evaluation of Yield-to-Pedestrian Channelizing Devices. Report FHWA-PA- 2006-010-050114. Pennsylvania Department of Transportation, Harrisburg. Tanaboriboon, Y., and J. Guyano. 1989. “Level-of-Service Standards for Pedestrian Facilities in Bangkok: A Case Study.” ITE Journal, Vol. 59, No. 11, pp. 39–41. Tanaboriboon, Y., S.S. Hwa, and C.H. Chor. 1986. “Pedestrian Characteristics Study in Singapore.” Journal of Transportation Engineering, Vol. 112, pp. 229–235. Tarawneh, M.S. 2001. “Evaluation of pedestrian speed in Jordan with investigation of some contributing factors.” Journal of Safety Research, Vol. 32, No. 2, pp. 229–236. Thomas, L., B. Lan, R.L. Sanders, A. Frackelton, S. Garder, and M. Hintze. 2017. “Changing the Future? Development and Application of Pedestrian Safety Performance Functions to Prioritize Locations in Seattle, Washington.” Transportation Research Record: Journal of the Transportation Research Board, No. 2659, pp. 212–223. Thomas, L., L. Sandt, C. Zegeer, W. Kumfer, K. Lang, B. Lan, Z. Horowitz, A. Butsick, J. Toole, and R.J. Schneider. 2018. Systemic Pedestrian Safety Analysis. Pre-publication draft of NCHRP Research Report 893. Transportation Research Board, Washington, DC. Thomas, L., N.J. Thirsk, and C. Zegeer. 2016. NCHRP Synthesis 498: Application of Pedestrian Crossing Treatments for Streets and Highways. Transportation Research Board of the National Academies, Washington, DC. Tokunaga, R., N. Takahashi, and M. Asano. 2004. “Conjoint Approach as Customer-Based Measurement for Winter Walkway Maintenance in the Snowiest City of Japan.” Transportation Research E-Circular E-063. Sixth International Symposium on Snow Removal and Ice Control Technology, Spokane, WA. Torbic, D., D. Harwood, C. Bokenkroger, R. Srinivasan, D. Carter, C. Zegeer, and C. Lyon. 2010. “Pedestrian Safety Prediction Methodology for Urban Signalized Intersections.” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, pp. 65–74. Transportation Research Board. 1980. Transportation Research Circular 212: Interim Materials on Highway Capacity. Washington, DC. Transportation Research Board. 1985. Special Report 209: Highway Capacity Manual. Washington, DC. Transportation Research Board. 2000. Highway Capacity Manual 2000. Washington, DC. Transportation Research Board. 2010. Highway Capacity Manual. Washington, DC. Transportation Research Board. 2016. Highway Capacity Manual, Sixth Edition: A Guide for Multimodal Mobility Analysis. Washington, DC. Turner, S. and P. Lasley. 2013. “Quality Counts for Pedestrians and Bicyclists.” Transportation Research Record: Journal of the Transportation Research Board, No. 2339, pp. 57–67.

228 Turner, S., I. Sener, M. Martin, L.D. White, S. Das, R. Hampshire, M. Colety, K. Fitzpatrick, and R. Wijesundera. 2018. Guide for Scalable Risk Assessment Methods for Pedestrians and Bicyclists. Report FHWA-SA-18-032. Federal Highway Administration, Washington, DC. Van Derlofske, J.F., P.R. Boyce, and C.H. Gilson. 2003. “Evaluation of In-Pavement, Flashing Warning Lights on Pedestrian Crosswalk Safety.” International Municipal Signal Association Journal, Vol. 41, No. 3, pp. 20, 54–61. Van Hengel, D. 2013. “Build It and They Will Yield: Effects of Median and Curb Extension Installations On Motorist Yield Compliance.” Paper 13-3084. Presented at the 92nd Annual Meeting of the Transportation Research Board, Washington, DC. Van Houten, R. 1988. “The Effects of Advance Stop Lines and Sign Prompts on Pedestrian Safety in Crosswalk on a Multilane Highway.” Journal of Applied Behavior Analysis, Vol. 21, 1988, pp. 245–251. Van Houten, R., A.R. Retting, C.M. Farmer, and J. Van Houten. 2000. “Field Evaluation of a Leading Pedestrian Interval Signal Phase at Three Urban Intersections.” Transportation Research Record, No. 1734, pp. 86–92. Van Houten, R., B. Huitema, and H. Manal. 2014. “An Analysis of the Effects of Installing Pedestrian Countdown Timers on the Incidence of Pedestrian Crashes in the City of Detroit, Michigan.” Paper 14-0227. Presented at the 93rd Annual Meeting of the Transportation Research Board, Washington, DC. Van Houten, R., R. Ellis, J. Sanda, and J. Kim. 2006. “Pedestrian Push-Button Confirmation Increases Call Button Usage and Compliance.” Transportation Research Record: Journal of the Transportation Research Board, No. 1982, pp. 99–103. Van Houten, R., K. Healey, J.E. Malenfant, and R.A. Retting. 1998. “Use of Signs and Symbols to Increase the Efficacy of Pedestrian Activated Flashing Beacons at Crosswalks.” Transportation Research Record, No. 1636, pp. 92–95. Van Houten, R., J. LaPlante, and T. Gustafson. 2012. Evaluating Pedestrian Safety Improvements. Report RC-1585. Michigan Department of Transportation, Lansing. Van Houten, R., R. Ellis, and E. Marmolejo. 2008. “Stutter-Flash Light Emitting-Diode Beacons to Increase Yielding to Pedestrians at Crosswalks.” Transportation Research Record: Journal of the Transportation Research Board, No. 2073, pp. 69–78. Vasudevan, V., S. Pulugurtha, S. Nambisan, and M. Dangeti. 2011. “Effectiveness of Signal-Based Countermeasures on Pedestrian Safety: Findings from a Pilot Study.” Transportation Research Record: Journal of the Transportation Research Board, No. 2264, pp. 44–53. Virkler, M. 1998. “Prediction and Measurement of Travel Time Along Pedestrian Routes.” Transportation Research Record, No. 1636, pp. 37–42. Virkler, M., and S. Elayadath. 1994. “Pedestrian Speed–Flow–Density Relationships.” Transportation Research Record, No. 1438, pp. 51–58. Virkler, M., S. Elayadath, and G. Saranathan. 1995. “High Volume Pedestrian Crosswalk Time Requirements.” Transportation Research Record, No. 1495, pp. 41–48. Wang, K., and M. Liao. 2012. “Modeling Pedestrian Degree of Satisfaction Based on Structure Equation Model.” Proceedings, International Conference of Logistics Engineering and Management, Chengdu, China. DOI: 10.1061/9780784412602.0069. Wang, W., P. Li, W. Wang, and M. Namgung. 2012. “Exploring Determinants of Pedestrians Satisfaction with Sidewalk Environments: Case Study in Korea.” Journal of Urban Planning and Development, Vol. 138, No. 2, pp. 166–172. Weidmann, U. 1993. Transporttechnik der Fussgänger: Transporttechnische Eigenschaften des Fussgängerverkehrs, Literaturauswertung (Transportation engineering properties of pedestrian traffic, literature analysis). IVT Series No. 90. Institut für Verkehrsplanung, Transporttechnik, Strassen- und Eisenbahnbau, Zurich, Switzerland. Wier, M., J. Weintraub, E.H. Humphreys, E. Seto, and R. Bhatia. 2009. “An area-level model of vehicle–pedestrian injury collisions with implications for land use and transportation planning.” Accident Analysis & Prevention, Vol. 41, No. 1, pp. 137–145. Willis, A., Gjersoe, N., Havard, C., Kerridge, J. and Kukla, R. 2004. “Human movement behavior in urban spaces: implications for the design and modelling of effective pedestrian environments.” Environment and Planning, Vol. 31B, pp. 805–828. Wilson, D.G., and G.B. Grayson. 1980. Age-Related Differences in the Road Crossing Behaviour of Adult Pedestrians. Report LR 933. Transport Research Laboratory, Berkshire, U.K. Yao, L., L.S. Sun, Z.Y. Zhang, S.W. Wang, and J. Rong. 2012. “Research on the behavior characteristics of pedestrian crowd weaving flow in transport terminal.” Mathematical Problems in Engineering, Article ID 264295.

229 Young, S.B. 1998. “Evaluation of Pedestrian Walking Speeds in Airport Terminals.” Transportation Research Record, No. 1674, pp. 20–26. Yuan, J., and D.M. Dulaski. 2017. “Drivers’ Yielding Behavior at Crosswalks with Varying Treatments During the Day and Night.” Paper 17-02995. Presented at the 96th Annual Meeting of the Transportation Research Board, Washington, DC. Zegeer, C., D. Carter, W. Hunter, J. Stewart, H. Huang, A. Do, and L. Sandt. 2006. “Index for Assessing Pedestrian Safety at Intersections.” Transportation Research Record: Journal of the Transportation Research Board, No. 1982, Washington, DC, pp. 76–83. Zegeer, C.V., K.S. Opiela, and M.J. Cynecki. 1985. Pedestrian Signalization Alternatives. Final Report. Federal Highway Administration, McLean, VA. Zegeer, C.V., C. Seiderman, P. Lagerway, M. Cynecki, M. Ronkin, and R. Schneider. 2002. Pedestrian Facilities Users Guide—Providing Safety and Mobility. Publication FHWA-RD-01-102. Federal Highway Administration, Washington, DC. Zegeer, C.V., R. Stewart, H. Huang, and P. Lagerwey. 2005. Safety Effects of Marked Versus Unmarked Crosswalks at Uncontrolled Locations: Final Report and Recommended Guidelines. Report FHWA–HRT–04–100. Federal Highway Administration, Washington, DC. Zegeer, C.V., D. Nabors, and P. Lagerway. 2013. Pedestrian Safety Guide and Countermeasure Selection System (PEDSAFE). Federal Highway Administration, Office of Safety, Washington, DC. http://pedbikesafe.org (As of Sept. 19, 2018). Zegeer, C.V., R. Srinivasan, B. Lan, D.L. Carter, S. Smith, C. Sundstrom, N.J. Thirsk, J. Zegeer, C.V. Lyon, E. Ferguson, and R. Van Houten. 2017a. “Development of Crash Modification Factors for Uncontrolled Pedestrian Crossing Treatments.” Transportation Research Record: Journal of the Transportation Research Board, No. 2636, pp. 1–8. Zegeer, C.V., R. Srinivasan, B. Lan, D.L. Carter, S. Smith, C. Sundstrom, N.J. Thirsk, J. Zegeer, C.V. Lyon, E. Ferguson, and R. Van Houten. 2017b. NCHRP Research Report 841: Development of Crash Modification Factors for Uncontrolled Pedestrian Crossing Treatments. Transportation Research Board, Washington, DC. Zhao, L., Y. Bian, J. Rong, X. Liu, and S. Shu. 2015. “Modeling Pedestrian Level of Service on Sidewalks with Multi- factors Based on Different Pedestrian Flow Rates.” Paper 15-2016. Presented at the 95th Annual Meeting of the Transportation Research Board, Washington, DC. Zuniga-Garcia, N., H.W. Ross, and R.B. Machemehl. 2018. “Multimodal Level of Service Methodologies: Evaluation of the Multimodal Performance of Arterial Corridors.” Transportation Research Record: Journal of the Transportation Research Board, DOI: 10.1177/0361198118776112.