Application of Pedestrian Crossing Treatments for Streets and Highways (2016)

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.

113 APPENDIX B Detailed Synthesis of Literature Review of Treatments Roadway Design Features The following tables provide summaries of individual studies as follows: Treatment descriptions, location types where studied, study type, and the measured safety effects, which may include crashes, or behavioral/operational measures such as motorist speed, motorist yielding to pedestrians in crosswalks, pedestrian delay, pedestrian compliance with crosswalks, and other safety indicators. Narrow Lane Width Narrow Lane Width Location Type Study Type Measured Safety Effects Citation 1- Projects that increased the total number of through lanes by narrowing lanes 2- Projects to provide space for installation of center two-way left-turn lane -Urban arterial streets with posted speed limits of 45 mph or less -Roads with two- lane undivided cross sections to roads with eight through lanes Observational evaluation of 35 improvement projects that used lane widths of 10 ft or less 1- Intersection crashes increased. No effect on seg- ment crashes. 2- Total crashes reduced by 24 to 53%. Harwood (1990) Cross-sectional study to identify potential predictors of 85th percentile operating speed 19 (of 24 initially planned) horizontal curve sites and 36 straight section sites located in several suburban areas of Texas Cross-sectional analysis The study attempted to determine explanatory cross-section, roadside, activity and access, and posted speed limit and signing effects on 85th per- centile speeds at horizontal curve and tangent mid- points. With posted speed limit allowed as a predic- tor in the models, speed limit was the only significant explanatory variable for operating speed at tangent sections. On curves, in addition to posted speed, deflection angle and access density class also helped predict operating speed. With speed limit not allowed in the models, lane width was the only significant predictor of speed on tangent sections, but explained much less of the variance in speed than did speed limit. On curve sections, when posted speed was not included, the effect of median presence and road- side development were the only other significant predictors of speed, and again, the model explained less of the speed variation than when posted speed limit was allowed in the model. Fitzpatrick et al. (2001) and final report to TxDOT Varied Varied Varied Speed measurements See Engineering Speed Management Counter- measures: A Desktop Ref- erence of Potential Effec- tiveness in Reducing Speed (2014) Reference for Narrow Lane Width Engineering Speed Management Countermeasures: A Desktop Reference of Potential Effectiveness in Reducing Crashes, Federal Highway Administration, Washington, D.C., July 2014 [Online]. Available: http:/safety.fhwa.dot.gov/speedmgt/ ref_mats/eng_count/2014/reducing_crashes.cfm. Fitzpatrick, K., P.J. Carlson, M.D. Wooldridge, and M.A. Brewer, “Design Factors That Affect Driver Speed on Suburban Streets,” Transportation Research Record: Journal of the Transportation Research Board, No. 1751, Transportation Research Board of the National Academies, Washington, D.C., 2001, pp. 18–25.

114 Harwood, D.W., NCHRP Report 330: Effective Utilization of Street Width on Urban Arterials, Transportation Research Board, National Research Council, Washington, D.C., 1990, 55 pp. Road Diet Road Diet Location Type Study Type Safety Effects Citation -Most sites had reductions from four to three lanes. -Some sites from four to two or six or four lanes -New York City Urban and Subur- ban; 460 road seg- ments and 324 adjacent intersec- tions with road diets -Both controlled and uncontrolled locations -5 years before, 2 years after road diet installations -Collision data collected for treatment and comparison sites. -Roadway Segments: ANCOVA adjusted decrease in pedestrian crashes of 62% (CMF of 0.38, 0.17 s.e.) at road segments. -Decrease in (all types) of injurious and fatal colli- sions at road segments (CMF of 0.27, 0.08 s.e.) -Intersections: Nonsignificant ANCOVA adjusted decreasing trend in total pedestrian crashes at treat- ment sites. -Decrease in (all types) injurious and fatal colli- sions of 73% (CMF of 0.27, 0.08 s.e.) at intersections Chen et al. (2013) Comparison of pedestrian safety and crashes at 2,000 uncontrolled intersections and midblock sites with marked and unmarked crossings; roads included four to eight lanes, medians and median islands, and no median islands Urban and subur- ban sites in 30 cit- ies across the U.S. Comparative study of 5 years of pedestrian crashes at 1,000 marked cross- walks and 1,000 matched unmarked com- parison sites Two and three lane roads were associated with reduced pedestrian crash risk when compared with roads with four or more lanes. Zegeer et al. (2001) Conversions from four- lane, undivided corridors to fewer motorized lanes and converting space to other uses such as center turn lanes, pedestrian refuges, bike lanes, or parking -Corridors in large urban areas -Rural highways passing through small urban areas Guide on road diet treatments and synthesis of multiple study results for acci- dent modifications -Road diet treatments can be expected to reduce total crashes by an average of 19% on corridors in large urban areas; and 47% for rural highways pass- ing through small urban areas. -Actual crash reductions vary depending on rural/ urban nature of area, crash history, and traffic volume. -Road diets with greater operating speed decreases have potential to lower crash severity. Knapp et al. (2014); Harkey et al. (2008) Reductions from four- to three-lane roads including reallocating road space to provide room for sidewalks and bike lanes -Simulated urban arterials with four to three lane reductions -Daily traffic vol- umes of 10,000, 15,000, 17,500, and 20,000 vpd were chosen for the sensitivity analysis. -These were con- verted to hourly volumes of 500, 750, 875, and 1,000 vehicles per hour per direction. Used simulation software and synthesis of other studies to conduct a sensi- tivity analysis to determine at what hourly vol- ume the arterial vehicle LOS would decline with addition of road diet treatment. -A two-way, peak hour volume of around 1,750 vpd (875 per direction) was the threshold when a decrease in motor vehicle LOS was observed. -This could be mitigated with signal timing optimization. Knapp and Giese (2001) References for Road Diet Chen, L., C. Chen, R. Ewing, C.E. McKnight, R. Srinivasan, and M. Roe, “Safety Countermeasures and Crash Reduction in New York City—Experience and Lessons Learned,” Accident Analysis & Prevention, Vol. 50, 2013, pp. 312–322. Harkey, D., R. Srinivasan, and J. Baek, NCHRP 617: Accident Modification Factors for Traffic Engineering and ITS Improvements, Transportation Research Board of the National Academies, Washington, D.C., 2008.

115 Knapp, K., B. Chandler, J. Atkinson, T. Welch, H. Rigdon, R. Retting, S. Meekins, E. Widstrand, and R.J. Porter, Road Diet Informational Guide, Report No. FHWA-SA-14-028, Office of Safety, Federal Highway Administration, Washington, D.C., 2014 [Online]. Available: http:/safety.fhwa.dot.gov/road_diets/info_guide/. Knapp, K.K. and K. Giese, Guidelines for the Conversion of Urban Four-Lane Undivided Roadways to Three-Lane Two- Way Left-Turn Lane Facilities, Final Report, Sponsored by Iowa Department of Transportation’s Office of Traffic and Safety, Center for Transportation Research and Education, Iowa State University, Ames, 2001. Zegeer, C.V., J.R. Stewart, H. Huang, and P. Lagerwey, “Safety Effects of Marked versus Unmarked Crosswalks at Uncontrolled Locations: Analysis of Pedestrian Crashes in 30 Cities (with Discussion and Closure),” Transportation Research Record: Journal of the Transportation Research Board, No. 1773, 2001, Transportation Research Board of the National Academies, Washington, D.C., pp. 56–68. Raised Medians and Pedestrian Median Islands Raised Medians/Median Islands Location Type Study Type Safety Effects Citation -Suburban: Study sites were four- to six-lane arte- rials with raised medians. Comparison sites were four- to six-lane arterials with center two-way, left- turn lanes, and two- to six- lane undivided arterials. -Central business district: Study sites were mostly six-lane arterials with raised medians. Compari- son sites were four- to six- lane arterials with two-way, left-turn lanes and undi- vided three-, four-, and six- lane arterials. -CBD and subur- ban areas in Atlanta, GA; Phoe- nix, AZ; and Los Angeles, CA -Traffic volumes ranged mostly from 20,000 to 50,000 vpd. -Cross-sectional with sepa- rate analysis performed for suburban area types and CBDs to account for different pedestrian volumes. -Compared group mean crash rates for pedestrian and vehi- cle crashes, using ANOVA. -Sample of 32,894 vehicle and 1,012 pedestrian crashes -Suburban Arterials: Pedestrian collision rates were significantly lower on arteri- als with raised medians compared with undivided arterials, and lower compared with arterials with two-way, left-turn lanes. Vehicle collisions were signifi- cantly lower on arterials with raised medians compared with both undivided arterials and those with two-way, left- turn lanes. -CBD Arterials: Pedestrian collision rates were significantly lower on arte- rials with raised medians or with two- way, left-turn lanes compared with undivided arterials. Vehicle crash rates were lowest on arterials with two-way, left-turn lanes. Bowman and Vecellio (1994) -Raised medians on roads ranging from four to eight lanes at 2,000 uncontrolled crossing sites -140 crossings with raised medians and marked cross- walks; 135 crossings with raised medians and unmarked crosswalks; 1,725 sites without raised medians (860 marked and 865 unmarked) -Urban and subur- ban areas in 28 cities -Daily counts at each site were expanded to esti- mated daily pedes- trian volumes. -Cross-sectional analysis of 5 years of pedestrian crash data -The study primarily aimed to assess the safety effects of marking crosswalks at uncontrolled locations, but was also able to examine effects of raised medians at both marked and unmarked locations. The study controlled for motor vehicle and pedestrian annual volumes (pedes- trian volumes were estimated from hourly counts, and expanded by daily counts at some locations) and suggests that pedestrian crash reductions in the range of 46% might be expected from adding raised medians to multilane roads (CMF = 0.54). Zegeer et al. (2002) Four refuge islands Two unsignalized, four-leg intersec- tions on urban and suburban residen- tial roads in Sacra- mento, CA Observational before and after analysis of behavioral effects -A slight but nonsignificant increase in the proportion of drivers who yielded to pedestrians from 33% to 42% -An increase in the proportion of pedestrians who crossed in the cross- walk when refuge islands were installed at two, four-leg, unsignalized intersections -No effect on crossing delay after ref- uge islands were installed at two unsignalized, four-leg intersections in an urban, residential area Huang and Cynecki (2000)

116 Raised Medians/Median Islands Location Type Study Type Safety Effects Citation Pedestrian refuge islands Two signalized intersections in urban San Fran- cisco, CA Observational before and after analysis of behavioral effects -No significant impacts of pedestrian refuges on driver yielding, on pedes- trian–vehicle conflicts, nor on the pro- portion of trapped pedestrians at two urban, signalized intersections -An increase in crossing delay at one of the two intersections studied Pécheux et al. (2009) Combinations of high-visi- bility crosswalks, median refuge, Danish offsets, and pedestrian channelization Eight sites on minor and major arterials in the Las Vegas Metro Area Observational before and after analysis of behavioral effects -On minor and major arterials when median refuges were part of the treat- ment package (eight locations in all; various treatment combinations) there were -Statistically significant increases in motorists yielding to pedestrians, and -Statistically significant increases in the proportion of pedestrians who looked for vehicles before crossing. Pulugartha et al. (2012) Curb extension, pedestrian refuge island, and stop bars were installed. One site at a multi- lane arterial in the coastal zone in Santa Barbara, CA Observational before and after analysis of behavioral effects -An increase in the proportion of driv- ers who yielded in the far lane, but not the near curb lane after refuge islands, curb extensions, and stop bars were installed (from 61.5% in the before condition, to 82% in the after condition) -No significant difference in yielding by near-lane drivers -Crossing delay decreased by a statis- tically significant average of 4.9 s. van Hengel (2013) Removable pedestrian islands and pedestrian crossing signs One rural, two-lane site in a county in Minnesota -Before and after -Free-flow speed data (although collected at differ- ent time periods) was col- lected about 2 weeks after installation and one set about 6 weeks after installation. -Significant reduction in mean speed -Improved driver compliance to speed limit, both in the short term and the long term at site with pedestrian islands and crossing signs Kamyab et al. (2003) Major reconstruction of roadway included a 7.5-ft- wide, 900-ft-long raised median between the two interior intersections; nar- rowing road width, adding new signals and intersec- tion designs, and adding curbs and sidewalks. Urban four-lane roadway with four major intersections at 1,000-ft intervals in New Jersey. Traffic volume was less than 10,000 ADT. A collision analysis project- ing future collision rates and costs including before and after analysis of behavioral effects A 2 mph decrease in 85th percentile speed resulting from the major recon- struction/redesign including raised median King et al. (2003) Pedestrian refuge islands 28 sites in urban and suburban Toronto, Canada 3 years before and 6 years after crash comparison, with no analysis control for vol- ume, time trends, severity, or other factors -Pedestrian collisions dropped from 22 in the 3-year period before installation to six collisions afterward. -There were 46 vehicle–island crashes during the after period. -The authors mentioned that design and lane alignment changes might be able to mitigate motorist crashing into island effects. Bacquie and Egan (2001)

117 References for Raised Medians and Pedestrian Median Islands Bacquie, R. and D. Egan, “Pedestrian Refuge Island Safety Audit,” presented at Improving Transportation Systems Safety and Performance, ITE Spring Conference and Exhibit, Monterey, Calif., 2001. Bowman, B.L. and R.L. Vecellio, “Effect of Urban and Suburban Median Types on Both Vehicular and Pedestrian Safety,” Transportation Research Record 1445, Transportation Research Board, National Research Council, Washington, D.C., 1994, pp. 169–179. Huang, H.F. and M.J. Cynecki, “Effects of Traffic Calming Measures on Pedestrian and Motorist Behavior,” Transportation Research Record 1705, Transportation Research Board, National Research Council, Washington, D.C., 2000, pp. 26–31. Kamyab, A., S. Andrle, D. Kroeger, and D.S. Heyer, “Methods to Reduce Traffic Speed in High-Pedestrian Rural Areas,” Transportation Research Record: Journal of the Transportation Research Board, No. 1828, Transportation Research Board of the National Academies, Washington, D.C., 2003, pp. 31–37. King, M. R., J.A. Carnegie, and R. Ewing, “Pedestrian Safety Through a Raised Median and Redesigned Intersections,” Transportation Research Record: Journal of the Transportation Research Board, No. 1828, Transportation Research Board of the National Academies, Washington, D.C., 2003, pp. 56–66. Pécheux, K., J. Bauer, and P. McLeod, Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reducing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures Final System Impact Report, ITS Joint Program Office, HVH-1, U.S. Department of Transportation, Washington, D.C., 2009, 117 pp. [Online]. Available: http:/ safety.fhwa.dot.gov/ped_bike/tools_solve/ped_scdproj/sys_impact_rpt/sys_impact_rpt.pdf. Pulugartha, S.S., V. Vasudevan, S.S. Nambisan, and M.R. Dangeti, “Evaluating the Effectiveness of Infrastructure-Based Countermeasures on Pedestrian Safety,” Transportation Research Record: Journal of the Transportation Research Board, No. 2299, Transportation Research Board of the National Academies, Washington, D.C., 2012. van Hengel, D., “Build It and They Will Yield: Effects of Median and Curb Extension Installations on Motorist Yield Compliance,” Transportation Research Board 92nd Annual Meeting Compendium of Papers, Paper No. 13-3084, Washington, D.C., 2013. Zegeer, C.V., R. Stewart, H. Huang, and P. Lagerwey, Safety Effects of Marked Versus Unmarked Crosswalks at Uncontrolled Locations: Executive Summary and Recommended Guidelines, Report FHWA-RD-01-075, Federal Highway Administration, McLean, Va., 2002, 33 pp. Raised Crosswalk/Speed Table Raised Crosswalk/Speed Table Location Type Study Type Safety Effects Citation Raised crosswalks, and one site that also had an over- head beacon that was previ- ously installed Three sites: two institutional streets (University Cam- pus) in Durham, NC, and one T-intersection in Montgomery County, MD, on a residential collec- tor street Comparative cross- sectional analysis of behavioral effects -Vehicle operating speeds (50th percentile) were from 4 to 12.4 mph lower at each of the Durham treatment sites com- pared with control sites (no before data were available) and also trended lower at the Maryland site. -Motorist yielding to pedestrians was also significantly higher at the North Carolina site with both a raised cross- walk and the flashing beacon compared with its control loca- tion, but the numbers of pedestrians didn’t allow firm con- clusions at the other location in North Carolina or in Maryland. Huang and Cynecki (2000) Speed humps installed just prior to two-way bicycle path crossings. Several approaches to two-way bicycle crossings in the city of Helsinki, Finland Observational study of driver scanning behavior while approaching bicycle crossings Speed humps caused drivers to slow and look in both direc- tions more often. Summala et al. (1996)

118 References for Raised Crosswalk/Speed Table Huang, H.F. and M.J. Cynecki, “Effects of Traffic Calming Measures on Pedestrian and Motorist Behavior,” Transporta- tion Research Record: Journal of the Transportation Research Board, No. 1705, Transportation Research Board, National Research Council, Washington, D.C., 2000, pp. 26–31. Summala, H., E. Pasanen, M. Raesaenen, and J. Sievaenen, “Bicycle Accidents and Drivers’ Visual Search at Left and Right Turns,” Accident Analysis & Prevention, Vol. 28, No. 2, 1996, pp. 147–153. Curb Extensions/Bulb-Outs Curb Extensions Location Type Study Type Safety Effects Citation Known as “neck-downs” in New York City, and installed at corners of inter- sections and midblock crossings since 1968 Six locations with similarly designed neck-downs around urban New York City, from Manhat- tan to Staten Island Analysis of 5 to 10 years of collision data across all six study sites. Treated sites were compared with nearby sites without neck-downs. Total of 204 pedestrian crashes. -Pedestrian crash data were still too limited to identify statis- tically significant pedestrian crash effects. -Researchers analyzed effects by each location separately, and the trends were mixed with regard to pedestrian crash frequency and severity and total crash frequency and severity. -There were indications that overall crash severity may be lower. King (1999) Curb extensions Eight residential and arterial crosswalk locations in Massa- chusetts, Washing- ton, North Caro- lina, and Virginia Before–after treat- ment (periods unre- ported) to measure behavioral effects No significant effects on motorist yielding, speed, or pedes- trian wait time could be detected. Huang and Cynecki (2000) Curb extension, pedestrian refuge island, and stop bars were installed. One site: a multi- lane arterial in the coastal zone in Santa Barbara, CA Before–after analysis of behavioral effects -Significant improvements in driver yielding in the far lane of the crossing for staged crossings (from 61.5% in the before condition, to 82% in the after condition) -No improvement was observed in the near lane. -Pedestrian wait times and the distance at which motorists yielded in advance of the crosswalk were also reported to improve by a statistically significant average of 4.9 s. van Hengel (2013) Uncontrolled marked crosswalk had a curb exten- sion on only one side, but advanced stop bars were at both sides of the two-lane street with average speeds exceeding the speed limit. One suburban intersection cross- walk in Albany, OR Traffic volume was 17,000 vpd. Observational. No data were collected prior to installation of curb extension; thus, the researchers used the untreated side of the same intersection as the control. They observed 99 crossings. -Significantly fewer vehicles passed before one yielded to pedestrians crossing on the curb extension side, indicating a safety and delay benefit. -Fewer vehicles stopped at the advance stop bar on the treat- ment side compared with the untreated side. Johnson (2005) References for Curb Extensions/Bulb-Outs Huang, H.F. and M.J. Cynecki, “Effects of Traffic Calming Measures on Pedestrian and Motorist Behavior,” Transportation Research Record 1705, Transportation Research Board, National Research Council, Washington, D.C., 2000, pp. 26–31. Johnson, R.S., Pedestrian Safety Impacts of Curb Extensions: A Case Study, Final Report FHWA-0R-DF-06-01, Oregon Department of Transportation, Salem, and Federal Highway Administration, Washington, D.C., 2005, 32 pp. King, M.R., “Calming New York City Intersections,” Transportation Research Circular E-C019: Urban Street Symposium Conference Proceedings, Dallas, Tex., June 28–30, 1999, I-3/ pp. 1–15.

119 van Hengel, D., “Build It and They Will Yield: Effects of Median and Curb Extension Installations on Motorist Yield Compli- ance,” Transportation Research Board 92nd Annual Meeting Compendium of Papers, Paper No. 13-3084, 2013. Corridorwide Speed Calming Corridorwide Speed Calming Location Type Study Type Safety Effects Citation Street narrowing, chokers, curb extensions, transverse rumble strips at the begin- ning of treated sections, cobblestones for short sec- tions, sometimes speed humps or other devices, traffic signals, parking Multiple cities that evaluated the effects of areawide urban traffic calm- ing measures, including on both local and main streets including traffic restricting/ diverting measures and traffic-calming measures Meta-analysis of 33 studies. One of the aims of the projects studied was to reclas- sify and restore the local functions and remove through traf- fic from residential areas and concentrate it on a few streets designated as main roads. Results of the meta-analysis indicated that these types of schemes reduced the number of total injury collisions for all road users by about 15%. Effects were greatest on residential streets (about 25%) and lower on main roads (10%) Elvik (2001) 12-ft humps, 14-ft humps, 22-ft humps, longer tables, raised intersections, circles, narrowing, one-lane slow points, half closures, diago- nal diverters Numerous projects throughout the United States An overview of pre- 1999 traffic-calming project evaluations carried out in the United States. Sample size, research design, data collection meth- ods, and durations were not standard- ized. Ewing cautions that the statistics be used as ballpark esti- mates of impact. 85th percentile downstream vehicle speeds (in mph) -12-ft humps: 179 sites: Average after calming = 27.4 mph, a reduction of 7.6 mph and 22% -14-ft humps: 15 sites: average after calming = 25.6 mph, a reduction of 7.7 mph and 23% -22-ft tables: 58 sites: average after calming = 30.1 mph, a reduction of 6.6 mph and 18% -Longer tables: 10 sites: average after calming = 31.6 mph, a reduction of 3.2 mph and 9% -Raised intersections: three sites: average after calming = 34.3 mph, a reduction of 0.3 mph and 1% -Circles: 45 sites: average after calming = 30.3 mph, a reduc- tion of 3.9 mph and 11%. -Narrowing: seven sites: average after calming = 32.3 mph, a reduction of 2.6 mph and 4% -One-lane slow points: five sites: average after calming = 28.6 mph, a reduction of 4.8 mph and 14% -Half-closures: 16 sites: average after calming = 26.3 mph, a reduction of 6.0 mph and 19% -Diagonal diverters: seven sites: average after calming = 27.9 mph, a reduction of 1.4 mph and 4% Ewing (1999, Table 5.1) Bulb-outs or curb exten- sions, raised intersections, and raised crosswalks Midblock and intersection loca- tions in seven states in the United States Study synthesis -Traffic-calming devices resulted in lower overall vehicle speeds. -Traffic-calming measures did not produce a statistically sig- nificant effect on average pedestrian waiting time. Huang and Cynecki (2000) Roundabout intersection designs, road diets, speed humps and tables, chicanes, and lateral shifts Numerous sites throughout United States Speed management action plan guidance and synthesis of evidence -Roundabout intersection designs, road diets, speed humps, and tables have the most evidence that they lower speeds and crashes. -Chicanes and lateral shifts have been found to lower speeds. Thomas et al. (2015, Speed Manage- ment Toolkit) References for Corridorwide Speed Calming Elvik, R., “Area-wide Urban Traffic Calming Schemes: A Meta-analysis of Safety Effects,” Accident Analysis & Prevention, Vol. 33, No. 3, 2001, pp. 327–336. Ewing, R.H., Traffic Calming: State of the Practice, Report FHWA-RD-99-135, Institute of Transportation Engineers, Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., 1999, 244 pp. [Online]. Available: http:/ www.ite.org/traffic/tcstate.asp#tcsop. Huang, H.F. and M.J. Cynecki, “Effects of Traffic Calming Measures on Pedestrian and Motorist Behavior,” Transportation Research Record 1705, Transportation Research Board, National Research Council, Washington, D.C., 2000, pp. 26–31.

120 Thomas, L., R. Srinivasan, M. Worth, M. Parker, and R. Miller, Jurisdiction Speed Management Action Plan Development Package, Report FHWA-SA-15-017, Federal Highway Administration, Washington, D.C., 2015 [Online]. Available: http:/ safety.fhwa.dot.gov/speedmgt/ref_mats/ Speed Management Toolkit available: http:/safety.fhwa.dot.gov/speedmgt/ref_ mats/docs/speedmanagementtoolkit_final.pdf. Pedestrian Overpass/Bridge and Underpass/Tunnel Pedestrian Overpass/ Underpass Location Type Study Type Safety Effects Citation Pedestrian overpass/bridge and underpass/ tunnel Tokyo, Japan, and San Francisco, CA Review of pedestrian safety research -An important measure of effectiveness for overpasses and underpasses is how much they are used by pedestrians. -“Measure of convenience (R), defined by the ratio of the time it took to cross the street on an overpass divided by the time it took to cross at street level.” -About 95% of pedestrians opt for the overpass if R = 1, meaning that it takes the same amount of time to cross using the overpass as it does at street level. If the overpass route takes 50% longer (R = 1.5), almost no one uses it. Under- passes were used even less often for similar values of R. Campbell et al. (2004) citing Moore and Older (1965); Mead et al. (2014) Pedestrian overpass 31 sites in Tokyo, Japan Review of pedestrian safety research citing a before and after comparison of pedes- trian crashes. Crashes occurring in 200-m (218-yard) and 100-m (109-yard) sections on either side of each site were tabulated. -After overpasses were installed, pedestrian crossing acci- dents decreased substantially, although nonrelated accidents increased by 23% in the 200-m sections. It is not known whether this increase could have been the result of other fac- tors unrelated to the overpass. -Daytime pedestrian crashes were reduced more than night- time crashes. This may be related to the volumes of pedestri- ans crossing the road. Campbell et al. (2004); Mead et al. (2014) Pedestrian overpass Three pedestrian overpasses in San Francisco Panel of disabled residents Potential hazards or barriers include inadequate or nonexis- tent railings on bridge approaches; steep cross slopes; lack of a level platform at the base on bridge ramps where wheel- chairs can stop before entering the street; inadequate sight distance to see opposing flow of pedestrians and a lack of level rest areas on spiral ramps; maze-like barriers on bridge approaches that are used to slow down bike traffic, but can also impede the progress of wheelchair-bound or visually impaired users; and lack of sound screening on the overpass so that the visually impaired can hear people coming the other way and avoid crashes. Campbell et al. (2004) citing Swan (1978); Mead et al. (2014) References for Pedestrian Overpass/Bride and Underpass/Tunnel Campbell, B.J., C.V. Zegeer, H.H. Huang, and M.J. Cynecki, A Review of Pedestrian Safety Research in the United States and Abroad, Report FHWA-RD-03-042, Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., 2004. Mead, J., C. Zegeer, and M. Bushell, “Evaluation of Pedestrian-Related Roadway Measures: A Summary of Available Research,” Pedestrian and Bicycle Information Center, Chapel Hill, N.C., 2014 [Online]. Available: http:/www.pedbikeinfo. org/cms/downloads/PedestrianLitReview_April2014.pdf.

121 Enhanced Illumination at Crossings Enhanced Illumination Location Type Study Type Safety Effects Citation Adding intersection illumination U.S. sites Meta-analysis on multiple studies -42% reduction in pedestrian nighttime crashes -This countermeasure is included in the Highway Safety Manual, with a caution that the adjusted standard error mea- surement is from 0.2 to 0.3. Elvik and Vaa (2004); CMF Clearing- house CMF/CRF Details (n.d.) Study of light conditions Midblock and intersection locations Analyzed 58,202 pedestrian crashes (less 23,634 removed because of data issues) to determine effects of light condi- tions on pedestrian crash severity Controlling for pedestrian and driver attributes, vehicle attri- butes, speed limit and roadway attributes, and weather con- ditions, street lighting reduced the odds of a fatal injury by 42% at midblock locations and 54% at intersections com- pared with dark conditions without street lighting. Chu (2006) Smart lighting system that used a pedestrian detection device in order to automati- cally increase illumination One midblock crosswalk in Las Vegas, NV. The site was chosen because the major- ity of motorists failed to yield to pedestrians and a high percentage of collisions occurred at night. Before–after. Data were collected before and after the treat- ment was installed at dawn and dusk hours, and included seven measures of effective- ness involving pedes- trian and driver behavior at the crosswalk. -A statistically significant increase in the percentage of diverted pedestrians (pedestrians who purposefully used the crosswalk) -A significant increase in the percentage of motorists yield- ing to pedestrians -A significant increase in the percentage of motorists yield- ing to pedestrians at greater than 10 ft before the crosswalk -Author concluded that the countermeasure helped to improve pedestrian safety, likely owing to the increased visi- bility and attention to pedestrians provided by enhanced lighting at the site. Nambisan et al. (2009) References for Enhanced Illumination at Crossings Chu, X., Pedestrian Safety at Midblock Locations, Report BD544-16, Florida Department of Transportation, University of South Florida Center for Urban Transportation Research, 2006 [Online]. Available: http:/www.dot.state.fl.us/research-center/ Completed_Proj/Summary_PL/FDOT_BD544_16_rpt.pdf. CMF Clearinghouse CMF/CRF Details, CMF ID: 436, Provide Intersection Illumination, Federal Highway Administra- tion, U.S. Department of Transportation, Washington, D.C. [Online]. Available: http:/www.cmfclearinghouse.org/detail. cfm?facid=436. Elvik, R. and T. Vaa, The Handbook of Road Safety Measures, 1st ed., Elsevier, Boston, Mass., 2004, 1078 pp. Nambisan, S.S., S.S. Pulugurtha, V. Vasudevan, M.R. Dangeti, and V. Virupaksha, “Effectiveness of Automatic Pedestrian Detection Device and Smart Lighting for Pedestrian Safety,” Transportation Research Record: Journal of the Transportation Research Board, No. 2140, Transportation Research Board of the National Academies, Washington, D.C., 2009, pp. 7–34.

122 Traffic Control Devices High-Visibility Crosswalks High-Visibility Crosswalk Markings Location Type Study Type Safety Effects Citation High-visibility crosswalk markings Urban—72 sites, New York City Two-group pretest/ posttest using ANCOVA modeling to account for regres- sion to the mean effect Before—5 years before crash data After—2 years after crash data After adjusting for comparison group (to account for poten- tial regression to the mean): -Statistically significant reduction of 48% in pedestrian crashes (CMF of 0.52; 0.17 s.e.) compared with if high-visi- bility crosswalks had not been added -No statistically significant effect on all crashes (0.99 CMF; 0.15 s.e.) or on multi-vehicle crashes (1.26 CMF; 0.20 s.e.) Chen et al. (2013) Two high-visibility types: continental markings and bar pairs University campus (Texas A & M) midblock crossings -Road “feel”— urban, mixed, and rural -30–45 mph posted limits -40–50 ft road width. -Various other con- ditions, including presence of side- walk (none, one side, two sides); number of lanes (two lanes, four lanes); with or without bike lanes; undivided or divided Comparison of visi- bility—through mea- sure of detection dis- tances of two types of high-visibility cross- walk markings (conti- nental, bar pairs) and standard markings (transverse lines), by 78 experimental sub- ject drivers of an instrumented vehi- cle—on approaches to crosswalk mark- ings newly installed at three locations each (nine total) on open roads on the Univer- sity of Texas Also obtained subject ratings of how easy it was to see each type of crosswalk -Detection distances for continental and bar pairs were sta- tistically similar and also statistically significantly longer than for transverse line markings at day and at night. -Detection distances for all marking types were also longer at daytime. -The presence of traffic also had the effect of reducing detec- tion distance. -Age, gender, driver eye height, and vehicle type were found to have minimal significance by the research team. -Participants also preferred the continental and bar pair markings to the transverse markings as reported in Iragavar- apu, Fitzpatrick, and Chrysler (2011) Fitzpatrick et al. (2011); Iragavar- apuet al. (2011) High-visibility crosswalks (yellow continental) Urban School Zones—San Fran- cisco, CA. 54 treated intersections. Included multiple lane sizes and speeds, controlled and uncontrolled, local (residential) and non-local streets, and a range of ADTs. Empirical Bayesian (EB) method, before- and-after collision analysis. Before—at least 5 years before crash data per site After—at least 3 years after crash data per site Sites installed between 1997 and 2003 54 untreated compari- son locations in geo- graphically close proximity and with similar ADTs Using the EB method to develop a safety performance func- tion for intersections without the treatment and comparing actual crashes with the treatment to expected crashes without the treatment, the treated effect was estimated to be: reduc- tion of 37% in pedestrian collisions (95% CI, 13% to 60%) Stated as a CMF, the effect is 0.63 (0.12 s.e.). Feldman et al. (2010)

123 High-Visibility Crosswalk Markings Location Type Study Type Safety Effects Citation Strong yellow green (SYG) crosswalk markings; SYG plus speed feedback: “The school safety program elements included school crossing warning signs on the approach to and at every intersection, 20 mph during School Hours When Children Are Present speed limit signs on every sur- rounding street, speed humps on the local streets around the school (at least in the 660-ft blocks), inter- active speed indicators on selected approaches along arterial street frontages, and strong SYG pavement markings at the speed humps, at all crosswalks, at ‘SCHOOL’ legends on arterial streets, and center lines and stop bars along streets adjacent to the schools.” Pavement mark- ings and signs Urban—15 sites with at least one arterial street front- age of the four blocks surrounding the school, 42–48 ft in width, with two-way, 24-hour volumes of 12,000– 20,000 vehicles per day, Chicago, IL -12 locations, crosswalk mark- ings only -3 locations, cross- walk markings plus speed feedback signs Comparison of traffic speeds during peak a.m. and p.m. school travel times, before and after crosswalks implemented; no comparison group -For all locations, the average mean speed (aggregated and weighted by volume) decreased by < 1 mph, from 26.3 mph before the installations, to 25.6 mph after (2.7%). -The percentage of traffic over the speed limit on arterials was 78.1% before and 77.5% after the installations, a reduc- tion of only 0.8%. -At locations with SYG markings plus speed feedback signs, the aggregate peak hour speed was 24.3 mph before the installation, and 25.0 mph after, an increase in speed of 2.9%. -The percentage of traffic going faster than the 20 mph school zone speed limit also increased by 2.9% after the installations. -For the 12 arterial segments with SYG and no feedback signs, the aggregate mean peak hour speed was 27.1 mph before the installations and 25.8 mph after, a decrease of 4.8%. -The percentage of traffic going faster than the 20 mph speed limit decreased slightly at arterial locations without feedback signs, from 79.4% before the installation to 77.5% after, a decrease of 2.4%. -On local streets (that typically also included traffic calming/ speed humps), speeds also increased slightly. The authors concluded that the SYG treatment and/or SYG plus speed feedback signs did not have a calming effect on speeds in school walk zones. However, they acknowledge the lack of comparison sites using traditional white markings. Chicago DOT (2005) High-visibility ladder style crosswalk plus novel over- head illuminated crosswalk sign. standard crossing signs, MUTCD W11A, and median islands were already present. Urban—two sites, Clearwater, FL T-intersections (stop-controlled on side streets) one 25 mph arterial street with beach traffic Cross-sectional com- parison of motorist yielding at treated locations compared with one location with standard pedes- trian signage (W11A) and crosswalk design, and one location with no crosswalk or warn- ing signs -Drivers found to be 30%–40% more likely to yield than drivers at the control locations. -At night, there was a smaller and nonsignificant difference in driver yielding of 8%. -Significantly higher percentage of pedestrians used the high-visibility crosswalks (35%) compared with others at similar locations. -There were no apparent differences in pedestrian overconfi- dence, running, or conflicts. Nitzburg and Knob- lauch (2001) References for High-Visibility Crosswalks Chen, L., C. Chen, R. Ewing, C. McKnight, R. Srinivasan, and M. Roe, “Safety Countermeasures and Crash Reduction in New York City: Experience and Lessons Learned,” Accident Analysis & Prevention, Vol. 50, 2013, pp. 312–322. Chicago Department of Transportation, Evaluation of School Traffic Safety Program Traffic Control Measure Effectiveness, 2005 [Online]. Available: http:/mutcd.fhwa.dot.gov/resources/policy/ygcrosswalkmarking/chicagostudy/yg_mkngs_chi- cago_rpt.pdf. Feldman, M., J.G. Manzi, and M.F. Mitman, “Empirical Bayesian Evaluation of Safety Effects of High-Visibility School (Yel- low) Crosswalks in San Francisco, California,” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, Transportation Research Board of the National Academies, Washington, D.C., 2010, (1), pp. 8–14. Fitzpatrick, K., S. Chrysler, V. Iragavarapu, and E. S. Park, “Detection Distances to Crosswalk Markings: Transverse Lines, Continental Markings, and Bar Pairs,” Transportation Research Record: Journal of the Transportation Research Board, No. 2250, Transportation Research Board of the National Academies, Washington, D.C., 2011, pp. 1–10. Also summarized in Fitzpatrick, K., S. Chrysler, R. Van Houten, W. Hunter, and S. Turner, 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, U.S. Department of Transpor- tation, Washington, D.C., 2011.

124 Iragavarapu, V., K. Fitzpatrick, and S.T. Chrysler, “Driver Preference for Crosswalk Marking Patterns,” Transportation Research Record: Journal of the Transportation Research Board, No. 2250, Transportation Research Board of the National Academies, Washington, D.C., 2011, pp. 57–64. Nitzburg, M. and R. Knoblauch, An Evaluation of High-Visibility Crosswalk Treatments—Clearwater, Florida, Report FHWA-RD-00-105, Federal Highway Administration, McLean, Va., 2001, 19 pp. [Online]. Available: http:/www.fhwa.dot. gov/publications/research/safety/pedbike/0105.pdf. Advanced Stop/Yield Bars/Signs Advance Stop/Yield Location Type Study Type Safety Effects Citation Crosswalk lines and removable advance stop lines and “Stop Here for Pedestrians” signs with an arrow pointing at an angle toward the road. Stop markings (an X) were also painted on the roadway 50 m in advance of the cross- walks on each side, but there were no traffic control devices at the crosswalk. Dartmouth, Nova Scotia. The first experiment was car- ried out at a six-lane crossing connecting two shopping malls; speed limit was 50 km/h (31 mph). The second experiment, conducted in the same manner 1 year later, reassessed behaviors at the first location (markings had been removed for 6 months), and added a crossing of a five-lane urban street con- necting a bus stop and resi- dential area. Observations were collected during day- light hours on week- days at pretreatment baseline, following treatment application, and following removal of the mark- ings and signs and reapplication two more times. -Although the percentage of motorists who yielded increased by only a slight amount, the introduction of the sign plus stop making condition resulted in a large increase in the percentage of motorists who stopped at least 10 ft from the crosswalk (from 50% to 95%) when they did stop. -The markings and signs produced an 80% decrease in vehicle–pedestrian conflicts. [During the second and third baseline conditions (signs and lines removed) conflicts increased again, but not as high as at initial baseline.] -Average pedestrian crashes and injuries also trended lower in the after period, but there were no controls for other potential causes. -Based on the results of this study, the Nova Scotia Department of Transportation began using advance yield markings throughout the province in order to mark crosswalks. Van Houten (1988) Advance stop lines and signs, education, and enforcement 34 uncontrolled crosswalks in three Canadian cities in the Newfoundland (NL) and New Brunswick (NB) provinces Before–after analysis of crash data and behavioral effects. B = 3.5 years; A= 2 years following the program. -Motorist yielding increased before to after in all three cities. Yielding increased from 54% to 81% in St. Johns (NL), from 9% to 68% in Fredericton (NB), and from 44% to 71% in Moncton–Dieppe (NB). -Given the treatment combination, it is unclear to what extent the advance stop lines and signs and the education and enforcement measures contrib- uted to the increase in motorist yielding behavior. -Average pedestrian crashes and injuries also trended lower in St. Johns and Fredericton and in the after period, but there were no controls for other potential causes. Malefant and Van Houten (1990) “Stop Here for Pedestrian” signs, followed by applica- tion of advance stop lines Two marked crosswalks in Dartmouth, Nova Scotia Pretreatment data were collected at baseline, and then fol- lowing the addition of the signs, and the cumulative effect of adding the stop lines 50 ft in advance of the crosswalks.q -The signs and advance stop lines had little effect on the percentage of motorists who yielded to pedestri- ans, but were associated with an increase in motorist yielding distance. -Pedestrian–vehicle conflicts decreased following installation of the stop lines. The baseline conflict rate before either treatment was 53%, decreasing to 10% at one location and to 6% at the second loca- tion following both treatments. Van Houten and Male- fant (1992) Advance yield markings (shark teeth) used with symbol signs Three crosswalks in Nova Scotia, Canada, where yellow flashing beacons were already in place: one midblock cross- walk spanning three lanes of traffic in each direction divided by a refuge island and two uncontrolled crossings of four-lane streets at T-intersec- tions with minor roads con- trolled by a stop sign Observational study of behavioral effects. The researchers experimented with yield marking placement. -Marking and sign placement were effective at dis- tances between 10 and 25 m in advance of the crosswalks. -Decreases in vehicle–pedestrian conflicts of 74%, 87%, and 67.1% at the three sites -A small increase in the percentage of motorists yielding, but higher percentages of motorists yielded at least 3 m in advance of the crosswalks Van Houten, et al. (2001)

125 Advance Stop/Yield Location Type Study Type Safety Effects Citation Flashing beacons. Advance yield line mark- ings with white back- ground “Yield Here to Pedestrians” signs, fluores- cent yellow-green “Yield Here to Pedestrian” signs (no markings), and fluores- cent yellow-green “Yield Here to Pedestrians” signs. 12 urban and 12 rural loca- tions throughout Nova Sco- tia, Canada, at uncontrolled approaches on multilane roads. Before–after study of behavioral effects. Once baseline data were collected for all 24 crosswalks, they were put into treat- ment groups of four, with one of the groups serving as a control throughout the experiment. Fol- low-up data were col- lected at 6 months following treatment installation. -Average number of vehicle–pedestrian conflicts decreased from 11.1% and 12.8% to 2.7% and 2.3%, respectively, at sites with the advance yield bar and either white or fluorescent sign. -No reduction of vehicle–pedestrian conflicts when the more conspicuous fluorescent yellow-green sign was used instead of the white sign. -Advanced stop lines with either type of sign were also associated with a statistically significant increase in motorist yielding from 69% to 85%. -The fluorescent yellow-green signs without advance yield bars were not associated with signifi- cant improvement in this study. The authors con- clude by recommending the installation of advanced yield markings 7 m to 18 m in advance of the crosswalk, in order to better increase pedes- trian visibility of oncoming vehicles when crossing. Van Houten et al. (2002) Combinations of advance yield markings and Danish offset, high-visibility cross- walk markings, and median refuge islands One unsignalized intersec- tion and one midblock loca- tion in Las Vegas, NV Cross-sectional observational study of driver and pedestrian behavioral effects. -Results indicated advance yield markings were more successful when coupled with Danish offsets (i.e., median crossing that turns to face oncoming traffic) as opposed to a combination with pedes- trian refuge islands. -A reduction in the number of pedestrians trapped in the roadway was observed at one of the study sites. -Both sites experienced an increase in pedestrians’ observing behavior. Nambisan et al. (2008) Stop lines and red visibility curb zones that prohibit on- street parking in the imme- diate vicinity of the intersection Two San Francisco intersections Part of review of mul- tiple studies on multi- ple countermeasures -No effects on driver yielding, vehicle stop posi- tion, and pedestrian–vehicle conflicts resulting from lines placed 4–10 ft before crosswalks. This distance may not have been noticeably different from the crosswalk lines to motorists. Pecheux et al. (2009) Curb extensions, pedestrian refuge island, advance stop lines-. One site in Santa Barbara, CA Before–after study of behavioral effects. Used staged cross- ings. Results were analyzed using cross tabulations and analy- sis of variance. -Crossing delay decreased by a statistically signifi- cant average of 4.9 s following the installation. -No significant difference in yielding of near-lane drivers was observed, but a statistically significant increase in yielding was observed for far-lane driv- ers, from 61.5% in the before condition, to 82% in the after condition. -A significant increase in motorist yielding distance van Hengel (2013) References for Advanced Stop/Yield Bars/Signs Malenfant, L. and R. Van Houten, “Increasing the Percentage of Drivers Yielding to Pedestrians in Three Canadian Cities with a Multifaceted Safety Program,” Health Education Research, Vol. 5, No. 2, 1990, pp. 275–279. Nambisan, S.S., V. Vasudevan, M. Dangeti, and V. Virupaksha, “Advanced Yield Markings and Pedestrian Safety: Analyses of Use with Danish Offsets and Median Refuge Islands,” Transportation Research Board 87th Annual Meeting Compendium of Papers, Paper No. 08-2994, 2008. Pécheux, K., J. Bauer, and P. McLeod, Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reduc- ing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures Final System Impact Report, ITS Joint Program Office, HVH-1, U.S. Department of Transportation, Washington, D.C., 2009, 117 pp. [Online]. Available: http:/safety.fhwa. dot.gov/ped_bike/tools_solve/ped_scdproj/sys_impact_rpt/sys_impact_rpt.pdf. van Hengel, D., “Build It and They Will Yield: Effects of Median and Curb Extension Installations on Motorist Yield Compliance,” Transportation Research Board 92nd Annual Meeting Compendium of Papers, Paper No. 13-3084, 2013. Van Houten, R., “The Effects of Advance Stop Lines and Sign Prompts on Pedestrian Safety in a Crosswalk on a Multilane Highway,” Journal of Applied Behavior Analysis, Vol. 21, No. 3, 1998, pp. 245–251.

126 Van Houten, R. and L. Malenfant, “The Influence of Signs Prompting Motorists to Yield Before Marked Crosswalks on Motor Vehicle-Pedestrian Conflicts at Crosswalks with Flashing Amber,” Accident Analysis & Prevention, Vol. 24, No. 3, 1992, pp. 217–225. Van Houten, R., J. Malenfant, and D. McCusker, “Advance Yield Markings: Reducing Motor Vehicle-Pedestrian Conflicts at Multilane Crosswalks with Uncontrolled Approach,” Transportation Research Record: Journal of the Transportation Research Board, No. 1773, Transportation Research Board National Research Council, Washington, D.C., 2 001, pp. 69–74. Van Houten, R., D. McCusker, S. Huybers, J. Malenfant, and D. Rice-Smith, “Advance Yield Markings and Fluorescent Yel- low-Green RA 4 Signs at Crosswalks with Uncontrolled Approaches,” Transportation Research Record: Journal of the Trans- portation Board, No. 1818, Transportation Research Board of the National Academies, Washington, D.C., 2002, pp. 119–124. In-Roadway “Yield to Pedestrians” Signs Advance Stop/Yield Location Type Study Type Safety Effects Citation In-roadway “Yield to Pedestrians” signs Three sites in first year, expanding to four sites in second year in Madison, WI 2-year evaluation of different forms of then-experimental in- street “Yield to pedestrians” signs -Motorist yielding increased at three of the four sites, but because of differences in the treatment applications, site geometrics, and pedestrian usage, the researchers could not draw further conclusions. Year 2 Field Eval- uation of Experimen- tal “In- Street” Yield to Pedestrian Signs (1999) Pedestrian safety cones with the message “State Law – Yield to Pedestrians in Crosswalks in Your Half of the Road” in middle of crosswalk Use of the cones was evalu- ated at six sites in New York State, and one site in Port- land, OR Before–after study of behavioral effects. Comparison with other treatments. -Pedestrian safety cones were the most successful in increasing the percentage of yielding drivers. Motorist yielding increased from 69.8% pretreat- ment to 81.2% posttreatment, which was signifi- cant at the 0.001 level. -Pedestrians who ran, aborted, or hesitated decreased, but the decrease was not statistically significant. The authors concluded that pedestrian safety cones were generally effective in increasing the percentages of pedestrians for whom motorists yielded. Huang et al. (2000) In-roadway “Yield to Pedestrians” signs Three sites in Cedar Rapids, IA, with traffic volumes from 5,400 to 25,000 ADT Before–after study of behavioral effects -While motorist yielding generally improved, and speeds were reduced for some locations and direc- tions, improvements were not uniform. -At one location, driver yielding from one direction increased only from 64% to 67% of the first vehi- cles arriving, whereas at the same crossing but opposite direction, yielding by the first arriving vehicle increased from 70% before to 84% after the treatment was installed. Speed effects by direction were also different at another location. Kannel et al. (2003) Removable pedestrian island and crossing signs combination Rural two-lane highway in Mahnomen County, MN Researchers collected pre- and posttreat- ment speed data to assess short- and long-term effects of the treatments. -A statistically significant reduction in mean speeds and a near doubling in speed limit compliance at the treatment site for both the long and short term in response to the combination of pedestrian island and pedestrian crossing signs. Kamyab et al. (2003)

127 Advance Stop/Yield Location Type Study Type Safety Effects Citation “Yield-to-Pedestrian” channelizing devices placed on the centerline of a roadway in advance of marked crosswalks Five locations each in four community types in Pennsyl- vania: urban, suburban, small city, and college town. Sites included both unsignal- ized intersections and mid- block locations. Before–after study of behavioral effects -Driver yielding increased from 30 to 34% at inter- section sites and from 17 to 24% at midblock crosswalks, and the effects did not appear to be related to community type. Pedestrians yielding to vehicles also decreased from 11 to 16% at intersec- tions, and from 8 to 13% at the midblock crossings. The researchers also found an increase in pedestri- ans using the crosswalks. The researchers con- cluded that the signs were more effective at inter- sections than midblock locations, but that they had positive effects at both location types. However, they also noted that follow-up data collection was complicated by damaged, moved, or missing signs. Strong and Kumar (2006) Impactable yield signs. All intersections are stop- controlled and have two- way flow. At one location, the signs were imple- mented at both a marked crosswalk and unmarked crosswalk of the same intersection. Installed at four crossings at three intersections in differ- ent areas of San Francisco. Each was a medium-sized, low-speed intersection located in an institutional, commercial, or industrial area. Street parking was present at all intersections. One T and two four-legged intersections were studied, with one of the four-leg sites also being a skewed intersection. Before–after study of behavioral effects -The only significant result that was common to all four sites was the increase in the rate of motorists yielding. At baseline, yielding rates varied from 20% to 40%, and at second follow-up, had increased to from 60% to 73% at all sites. -No significant effect on conflicts between drivers and crossing pedestrians, and there were no consis- tent effects on the other measures of effectiveness, including waiting time for pedestrians, percent of trips in the crosswalk, or time taken by pedestrians to cross a given crosswalk. The researchers cau- tioned that conflicts may be the best proxy for crash or safety risk, and that crash effects are the ultimate measure of effectiveness. Banerjee and Rag- land (2007) In-roadway “Yield to Pedestrians” signs Three crossings along the same street in Miami, FL. Carried an average of 29,500 vpd. Studied the effects of placing the “Yield to Pedestrians” signs at different distances from the crosswalk— at the crosswalk, and 20 ft in advance, 40 ft in advance, and all three locations at the same time -Test results found significantly more yielding at all three locations and that the signs placed at the crosswalk line, 20 ft in advance of the crosswalk line, 40 ft in advance of the crosswalk line, or at all three locations at the same time did not result in significantly widely different levels of yielding. -At location one, motorist yielding increased from 32% at baseline, to 70% when 40 ft away, 75% when 20 ft away, 78% at the crosswalk, and 79% when applied at all three locations. Similar results were found at the second location: before, 21% of motorists yielded improving to 54% at 40 ft, 63% at 20 ft, 65% when signs were placed at crosswalk line, and 56% when signs were at all three loca- tions. At the third location, 34% of motorists yielded increased to 69% with signs at the cross- walk line; 43%, 20 ft distant; 43%, 40 ft distant; and 52% with signs at all three locations. Ellis et al. (2007) Signs used in gateway configurations Sites in East Lansing, MI. Two study sites. The gateway configuration consists of one sign in the middle of the roadway, and two signs installed in the gutter pans on each side of the roadway. Staged crossings with three conditions were alternated and evalu- ated at the two study sites: no in-street sign (baseline), one in- street sign in the median (typical con- figuration), and three in-street signs in the gateway configuration. -At both sites, motorist yielding averaged 25% when no signs were present. The presence of one in-street sign was associated with motorist yielding of 57% at both locations. -The gateway (three signs) configuration was asso- ciated with 79% and 82% at the two locations. Thus, the gateway configuration using three signs was associated with the highest motorist yielding rates. Bennett et al. (2014)

128 Advance Stop/Yield Location Type Study Type Safety Effects Citation In combination with PHB Same study as above, but at two sites at suburban and urban locations in Detroit that had a PHB and in-street sign Same as above, but with PHB comparison At the suburban site, motorist yielding was 1% in the crosswalk-only condition. The addition of one in-street sign was associated with motorist yielding of 37%, and the use of the gateway treatment was associated with motorist yielding of 72%. This compared favorably to the PHB installed at the site. When the PHB was activated, motorists yielded 62% of the time. When one in-street sign was used along with the PHB at this location, motorist yielding increased to 85%. At the more urban location, an average of 10% of motorists yielded at baseline conditions. PHB activation was associated with a motorist yielding rate of 84%, and the addition of an in-street sign brought motor- ist yielding to 94.5%. Bennett et al. (2014) In-roadway “Yield to Pedestrians” signs Three university sites in Grand Forks, ND Observational behav- ioral effects Yielding was between 62% and 86% when the sign was not present, and between 72% and 98% when the signs were present (all locations considered). Yielding was higher when the signs were placed at the crosswalk (between 97% and 98%) compared with when it was placed further away (between 72% and 98%). Data collected about motorist speed indicated that minimum and maximum motorist speed was higher when no sign was pres- ent than when any sign was present at all but one of the eight studied locations. Gedafa et al. (2014) References for In-Roadway “Yield to Pedestrians” Signs Banerjee, I. and D.R. Ragland, Evaluation of Countermeasures: A Study on the Effect of Impactable Yield Signs Installed at Intersections in San Francisco, Institute of Transportation Studies, University of California, Berkeley, 2007 [Online]. Avail- able: http:/econpapers.repec.org/paper/cdlitsrrp/qt61k2k9bz.htm. Bennett, M.K., H. Manal, and R. Van Houten, “A Comparison of Gateway In-Street Sign Treatment to Other Driver Prompts to Increase Yielding to Pedestrians at Crosswalks,” Transportation Research Board 93rd Annual Meeting Compendium of Papers, Paper No. 14-0222, Washington, D.C., 2014. Ellis, R., R. Houten, and J. Kim, “In-Roadway ‘Yield to Pedestrians’ Signs: Placement Distance and Motorist Yielding,” Transportation Research Record: Journal of the Transportation Research Board, No. 2002, Transportation Research Board of the National Academies, Washington, D.C., 2007, pp. 84-89. Gedafa, D.S., B. Kaemingk, B. Mager,, J. Pape, M. Tupa, and T. Bohan, “Impacts of Alternative Yield Sign Placement on Pedestrian Safety,” Transportation Research Board 93rd Annual Meeting Compendium of Papers, Paper No. 14-1155, 2014. Huang, H., C. Zegeer, R. Nassi, and B. Fairfax, The Effects of Innovative Pedestrian Signs at Unsignalized Locations: A Tale of Three Treatments, Report FHWA-RD-00-098, Federal Highway Administration, McLean, Va., 2000, 31 pp. Kamyab, A., S. Andrle, D. Kroeger, and D.S. Heyer, “Methods to Reduce Traffic Speed in High-Pedestrian Rural Areas,” Transportation Research Record: Journal of the Transportation Research Board, No. 1828, Transportation Research Board of the National Academies, Washington, D.C., 2003, pp. 31–37. Kannel, E.J., R.R. Souleyrette, and R. Tenges, In-Street Yield to Pedestrian Sign Application in Cedar Rapids, Iowa, Final Report, CTRE Project 02-115, Office of Traffic and Safety, Iowa Department of Transportation, Ames, 2003 [Online]. Available: http:/www.intrans.iastate.edu/reports/pedyield.pdf. Strong, C. and M. Kumar, Safety Evaluation of Yield-to-Pedestrian Channelizing Devices, Pennsylvania Department of Transportation, Harrisburg, 2006. Year 2 Field Evaluation of Experimental “In-Street” Yield to Pedestrian Signs, Traffic Engineering Division, City of Madison Department of Transportation, Madison, Wis., 1999 [Online]. Available: http:/trid.trb.org/view/1999/M/748053.

129 Pedestrian Warning Signs Ped Warning Signs Treatments Location Type Study Type Safety Effects Citation Pedestrian warning signs Sites in central North Carolina Pre/post-treatment study of observational behavioral effects The use of this new sign was associated with increased numbers of cars that slowed down or stopped for pedestrians, although there was no decrease in conflict events following sign installation Clark et al. (1996) Pedestrian warning signs One site with two sign designs: the first was a picto- graph of a walking pedes- trian, which was added to a pedestrian-activated amber flashing beacon suspended over the roadway at the crossing site. It was coupled with a “Yield When Flash- ing” sign placed 50 m ahead of the crosswalk. Experimental: Evalu- ated the effects of two types of experimental signs on motorist yielding behavior Results indicated that both measures were effective in increasing motorist yield percentage, with the most effective treatment being the combination of the two. Only the “Yield When Flashing” sign was effective in reducing vehicle–pedestrian conflicts; the researchers theorized it was a result of the sign’s placement within adequate stopping distance of the crosswalk. Van Houten et al. (1998) Internally illuminated over- head crosswalk signs installed in conjunction with high-visibility crosswalks Two midblock crossing loca- tions in Clearwater, FL Using case-control research design, they compared motorist and pedestrian behav- ior at the treatment sites with two similar sites, one that fea- tured standard pedes- trian crossing signage and crosswalk design, and one that had no crosswalk. -The researchers found that during the day, drivers at the experimental crossing locations were 30%– 40% more likely to yield than drivers at the control locations. At night, there was a smaller and statisti- cally insignificant increase in driver yielding of 8%. -There was a significant increase in pedestrians using the crosswalk at the treatment sites compared with control sites, although the individual effects of having the signs in place could not be analyzed separately from the high-visibility crosswalk. Nitzburg and Knob- lauch (2001) Overhead yellow crosswalk sign and sign with amber lights around the edge that were activated by a pedes- trian pushbutton One intersection in Seattle, WA, and one in Tucson, AZ Before–after study of driver yielding -For the overhead sign: an increase in driver yield- ing from 45.5% before installation to 52.1%, which was significant at the 0.06 level. Following the installation of the sign, there was a statistically sig- nificant decrease in the percentage of pedestrians who ran, aborted, or hesitated in crossing. The researchers concluded that the overhead crosswalk sign was effective at encouraging driver yielding behavior. -For the amber lighted sign, the percentage of motorists yielding decreased in response to this sign. Neither of the two treatments led to a statisti- cally significant increase in crosswalk use; how- ever, the authors concluded that the overhead crosswalk sign and pedestrian safety cones were generally effective in increasing the percentages of pedestrians for whom motorists yielded. They cau- tioned that site characteristics would need to be taken into account when choosing or designing treatments to draw motorists’ attention to pedestri- ans in crosswalks. Huang et al. (2000) References for Pedestrian Warning Signs Clark, K.L., J.E. Hummer, and N. Dutt, “Field Evaluation of Fluorescent Strong Yellow-Green Pedestrian Warning Signs,” Transportation Research Record 1538, Transportation Research Board, National Research Council, Washington, D.C., 1996, pp. 39–46. Huang, H., C. Zegeer, R. Nassi, and B. Fairfax, The Effects of Innovative Pedestrian Signs at Unsignalized Locations: A Tale of Three Treatments, Report FHWA-RD-00-098, Federal Highway Administration, McLean, Va., 2000, 31 pp. Nitzburg, M. and R. Knoblauch, An Evaluation of High-Visibility Crosswalk Treatments—Clearwater, Florida, Report FHWA-RD-00-105, Office of Safety R&D, Federal Highway Administration, McLean, Va., 2001, 19 pp.

130 Van Houten, R., K. Healey, J.E.L. Malenfant, and R. Retting, “Use of Signs and Symbols to Increase the Efficacy of Pedes- trian-Activated Flashing Beacons at Crosswalks,” Transportation Research Record 1636, Transportation Research Board, National Research Council, Washington, D.C., 1998, pp. 92–95. In-Pavement Flashing Lights/Flashing Crosswalks In-Pavement Flashing Lights Treatment Location Type Study Type Safety Effects Citation In-pavement flashing lights Multiple cities Evaluation of many studies of the in- pavement flashing crosswalks -Short-term improvements in motorist yielding have been reported from most sites studied, but the addition of flashing lights alone may be insufficient to bring motorist yielding rates to a desirably high level—both daytime and nighttime rates often remained well below 50%. -Improvements at most locations were generally greater for nighttime when data were collected at both day and night, but yielding rates at night were also often much poorer, initially. -Initial gains in yielding rates also tended to decline over time and effects on motorist speed have varied across different studies. -Reported yielding rates also tended to be higher when observations included pedestrians already in their crossing or stepping out, compared with wait- ing to cross. Motorist yielding rates to pedestrians midway across also varied widely from site to site, and also tended to decline from initial improve- ments over a longer time period. Thomas (2006) In-pavement flashing lights Multiple cities Evaluation of many studies of the in- pavement flashing crosswalks -Only one study from a higher-volume arterial in Honolulu, HI, assessed impacts on pedestrian delay and motorized traffic flow. Pedestrians activated the flashing lights to create a gap for crossing, when in the before period, they waited for a gap to cross, thus helping reduce the delay to pedestrians. If pedestrians were frequent enough, this could potentially induce a greater impact to traffic flows than some other types of control. -Proportions of pedestrians using the crosswalks improved at some locations and not at others. The crosswalk location and convenience to desired routes likely has a bearing on this result, in addi- tion to any perceptions of enhanced safety. Thomas (2006) In-pavement flashing lights Sites in Vermont Observational field study -Found that while the system installed increased yielding and decreased approached speeds, they were removed due to multiple malfunctions and damage during winter (before an evaluation could be completed); DOT’s view that they preferred a treatment that would be visible during all seasons. Kipp and Fitch (2011) References for In-Pavement Flashing Lights/Flashing Crosswalks Kipp, W.M. and J. Fitch, “Evaluation of SmartStud™ In-Pavement Crosswalk Lighting System and BlinkerSign®,” Report No. 2011-3, Division Office, Federal Highway Administration, Montpelier, Vt., 2011, 17 pp. [Online]. Available: http:/ vtransplanning.vermont.gov/sites/aot_program_development/files/documents/materialsandresearch/completedprojects/ Evaluation_of_SmartStud_In_Pavement_Crosswalk_web.pdf. Thomas, L., “Safety Effects of In-Roadway Warning Lights or ‘Flashing Crosswalk’ Treatment: A Review and Synthesis of Research,” Pedestrian and Bicycle Information Center and FHWA, Chapel Hill, N.C., 2006 [Online]. Available: http:/www. pedbikeinfo.org/cms/downloads/FlashingCrosswalksReview_Thomas.pdf.

131 Overhead or Roadside-Mounted Flashing Beacons Overhead/Roadside Beacons Location Type Study Type Safety Effects Citation Pedestrian-activated flash- ing beacons Two crosswalks; urban and suburban sites in Dartmouth, Nova Scotia Experimental: two strategies for increasing the percentage of motorists yielding to pedestrians at crosswalks equipped with pedestrian- activated flashing beacons. One strategy involved add- ing an illuminated sign, with the standard pedestrian sym- bol next to the beacons. The second strategy involved placing signs 50 m before the crosswalk that displayed the pedestrian symbol and requested motorists to yield when the beacons were flashing. Both interventions increased yielding behavior and the effect of both together was greater than either alone. However, only the sign requesting motorists to yield when the beacons were flashing was effec- tive in reducing motor vehicle–pedestrian conflicts. Van Houten et al. (1998) One with a pedestrian pushbutton, and the other equipped with infrared sen- sors to automatically detect pedestrians Two types of flashing beacons at two sites were evaluated as part of a study of 13 treat- ments deployed at 29 sites throughout San Francisco. Before–after study: before 1–4 years; after 1–2 years -At one of the sites, the percentage of diverted pedestrians saw a statistically significant decrease following treatment, while the other site saw no impact. The same site measured a statistically sig- nificant decrease in trapped pedestrians, while the other site saw no impact. At both sites, there was a statistically significant increase in the percentage of yielding drivers. Both average pedestrian delay and pedestrian–vehicle conflicts decreased signifi- cantly at both sites. -It is important to note that both sites also had advance stop lines and one of the sites had an in- street “Yield to Pedestrians” sign. Hua at al. (2009) Overhead or roadside- mounted flashing beacons One site in Las Vegas, NV, with a flashing yellow signal at a midblock location with several other countermeasures: a high-visibility cross- walk, Danish offset, median refuge, and advanced yield markings 13 pedestrian and driver measures of effectiveness were studied by field observ- ers before and after the installation of the call button and analyzed using a two- proportion z-test. There was a significant reduction in the percentage of drivers blocking the crosswalks, as well as a sig- nificant increase in driver yielding distance in all distance categories (<10, 10–20, and >20 ft); how- ever, there was no statistically significant increase in the percentage of drivers yielding to pedestrians. Several pedestrian-related outcomes were also observed: an increase in the number of pedestrians looking for vehicles while crossing both halves of the street and an increase in the number of pedes- trians diverted to use the crosswalk. Vasudevan et al. (2011) Internally illuminated (not flashing beacon) overhead crosswalk signs that were installed in conjunction with high-visibility crosswalks Two midblock cross- ing locations in Clearwater, FL Using case-control research design, they compared motorist and pedestrian behavior at the treatment sites with two similar sites, one that featured standard pedestrian crossing signage and crosswalk design, and one that had no crosswalk. -During the day, drivers at the experimental cross- ing locations were 30%–40% more likely to yield than drivers at the control locations. At night, there was a smaller and statistically insignificant increase in driver yielding of 8%. There was a significant increase in pedestrians using the crosswalk at the treatment sites compared with control sites. -Although the individual effects of having the signs in place could not be analyzed separately from the high-visibility crosswalk, the researchers con- cluded that the treatments had a positive effect on pedestrian safety at the two intersections that were studied. Nitzburg and Knob- lauch (2001)

132 References for Overhead or Roadside-Mounted Flashing Beacons Hua, J., N. Gutierrez, I. Banerjee, F. Markowitz, and D.R. Ragland, PedSafe II Project Outcomes and Lessons Learned, Safe Transportation Research & Education Center, San Francisco 2009. Nitzburg, M. and R. Knoblauch, “An Evaluation of High-Visibility Crosswalk Treatments—Clearwater, Florida,” Publication FHWA-RD-00-105, Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., 2001. Van Houten, R., K. Healey, J.E. Louis Malenfant, and R. Retting, “Use of Signs and Symbols to Increase the Efficacy of Pedes- trian-Activated Flashing Beacons at Crosswalks,” Transportation Research Record: Journal of the Transportation Research Board, No. 1636(1), Transportation Research Board, National Research Council, Washington, D.C., 1998, pp. 92–95. Vasudevan, V., S.S. Pulugurtha, S.S. Nambisan, and M.R. Dangeti, “Effectiveness of Signal-Based Countermeasures for Pedestrian Safety: Findings from a Pilot Study,” Transportation Research Record: Journal of the Transportation Research Board, No. 2264, Transportation Research Board of the National Academies, Washington, D.C., 2011, pp. 44–53. Rectangular Rapid Flash Beacon (RRFB) RRFB Location Type Study Type Safety Effects Citation Stutter-flash LED beacons RRFBs were installed at two Miami-Dade County, FL, multi- lane crosswalks. Baseline data were collected pretreatment, and during the posttreatment phase the researchers alternated the acti- vation of the beacons at the sites in order to take further control measurements. Observ- ers measured the numbers of yielding motorists, vehicle– pedestrian conflicts, trapped pedestrians, and motorist yield- ing distance. -At the two sites, motorists yielding to resident pedestri- ans increased from 0% and 1% to 65% and 92%, respectively. -There was also a reduction in the number of vehicle– pedestrian conflicts and trapped pedestrians, leading the authors to conclude that the stutter-flash beacon was effective in increasing pedestrian safety at multilane crosswalks. Van Houten et al. (2008) RRFB Evaluation of RRFBs at two sites in Miami, FL. Traffic volume was 25,215 and 38,996. Before–after study (time period unknown). Assessed pedestrian and driver behavioral effects. -Statistically significant improvements in all of the stud- ied MOEs -Driver yielding for daytime and nighttime staged pedes- trian crossings increased from 4.2% before to 55.2% after at one location, and from 4.1% before to 60.1% at the second location. -Yielding to local resident pedestrians went from 12.5% and 5.4% to 73.7% and 83.4%, respectively. -The number of conflicts and pedestrians trapped in the roadway also declined significantly from 44% to 0.5% at one location. The percentage of conflicts also declined at both locations. Pecheux et al. (2009) One of the systems was pedestrian-acti- vated, while the sec- ond used an auto- mated pedestrian detection system. Two types of flash- ing beacons at two sites were evalu- ated as part of a study of 13 treat- ments deployed at 29 sites throughout San Francisco. Before–after study: before, 1–4 years; after, 1–2 years -The pushbutton-activated beacon led to a significant reduction in vehicle–pedestrian conflicts, from 6.7% pre- treatment to 1.9% posttreatment, as well as a significant increase in vehicles yielding, from 70% pretreatment to 80% posttreatment. -Only 17% of pedestrians activated the beacon, although an additional 27% of pedestrians crossed when the bea- con (47% total) was activated. -The automated flashing beacon led to a significant reduc- tion in vehicle–pedestrian conflicts (from 6.1% pretreat- ment to 2.9% posttreatment), a significant reduction in the number of trapped pedestrians (from 4.1% pretreatment to 0% posttreatment), and a significant increase in vehicle yielding (from 82 % pretreatment to 94% posttreatment. -Of the 13 countermeasures tested, both the pushbutton and automated flashing beacons were among the six countermeasures from this study that were considered the most effective in increasing pedestrian safety. Hua at al. (2009)

133 RRFB Location Type Study Type Safety Effects Citation RRFB One four-lane site at the intersection of Pinellas trail crossing in St. Petersburg, FL. Traffic volume 15,000 ADT. Before–after using video surveillance of behav- ioral effects. Data were collected on more than 1,000 pedestrians and bicyclists during daylight hours for 7 to 10 days during the before and after periods. -There were benefits to trail users (a majority of which were bicyclists) with a significant reduction in crossing delay, and a statistically significant increase in motorist yielding—from 2% pretreatment to 35% posttreatment, overall, and 54% when the beacon was activated. -There was also an improvement in the ability of pedes- trians to cross the entire intersection (from 82% pretreat- ment to 94% posttreatment). Hunter et al. (2009) RRFB 22 multilane, uncon- trolled crosswalks—19 in St. Petersburg, Florida, one in Washington, DC, and two in Mundelein, IL. 19 had speed limits of 35 mph, two of 30 mph, and one of 40 mph. ADTs ranged from < 5,000 vpd to 30,000 at the DC location, which also had a 30 mph posted limit. The researchers mea- sured driver yielding behavior, pedestrian– vehicle conflicts at base- line (pretreatment) and compared it with post- treatment data collected at eight times over the following two years to assess long-term effects. They were also compared with side and overhead beacons. -Yielding increased rapidly following installation across all sites, from an average 4% of drivers yielding before treatment, to an average of 84% at two years after instal- lation, demonstrating the longevity of effect. The range across all sites was from 72% to 96%. -Compared with standard overhead beacons: Motorist yielding increased from 15.5% with the overhead beacon to 78.3% when two RRFBs were installed, and increased further to 88% when four RRFBs were installed. -At the site of the side-mounted beacon, motorist yielding increased from 12% with the side-mounted beacon, to 72% with the installation of two RRFBs. -Data collected at night showed an increase in driver yielding behavior from 4.8% pretreatment to 84.6% (two- beacon RRFB) and 99.5% (four-beacon) posttreatment. Shurbutt and Van Houten (2010) RRFB Two Bend, OR, cross- walks at intersections along a 45-mph four- lane, bypass, limited access highway with a center median, bike lanes, and sidewalks. The 85th percentile speeds (before RRFBs were installed) were 54 mph and 52 mph. One crosswalk was at an intersection in the City of Bend—at a suburban to urban transition zone— on a four-lane facility with a center median. The posted speed limit is 35 mph; the 85th percen- tile speed was 44 mph. Before–after study (time periods unknown) of driver yielding. -Prior to the installation of the RRFBs, average motorist yield rates were 18% across all three locations. (Average for each site ranged from 5.8% to 24.7% at the crossings before RRFB.) -These rates increased to an 80% average for all three sites following treatment. (Average range for individual sites was from 74.3 to 82.9%.) Ross et al. (2011) RRFB Urban and suburban arte- rials in Portland, OR. Two sites: site 1 a five- lane, 35-mph urban arte- rial, with 30,700 vpd, and a narrow median refuge; and site 2 a TWLTL lane, 40-mph, suburban arte- rial, with 26,400 vpd, a median island, and a Danish offset crossing island. Reported on motorist yielding and pedestrian use of RRFBs when they were activated by the pedestrians versus not activated. -Site 1: Motorist yielding was 92% when beacons were activated and 75% when beacons were not activated. -Site 2: Motorist yielding was 91% when the RRFBs were activated and 45% when not activated. -Pedestrians activated the beacon 94% of the time at site 1 and 83% of the time at site 2. -Researchers also documented that motorists yielded more often to pedestrians in the second stage of their crossing at both locations. -At site 2 (the Danish offset), 82% of pedestrians who crossed the roadway chose to use the crosswalk, which compared favorably with a 71% compliance rate for marked midblock crosswalks in general. Foster et al. (2014)

134 RRFB Location Type Study Type Safety Effects Citation RRFB Six locations in Calgary, Canada. Traffic volumes ranged from 4,800 to 14,600 vpd, and streets had from one to five lanes. Observations of driver yielding using staged crossings The RRFBs were observed to increase yielding by an average of 15%, to nearly 100% motorist yielding at the majority of the study sites. Overall, the average motorist yielding increased from 83% to 98% following the instal- lation of the RRFB, indicating that the RRFBs were effective at increasing pedestrian safety at these six sites. Domarad et al. (2013) RRFB with school zone signs Multilane crossings in Texas with PHBs, RRFBs, and traffic con- trol signals (TCS) installed. 22 RRFB sites in three different Texas cities. Posted speed limits ranged from 30 mph (two sites in one city) to 45 mph (two sites in differ- ent cities) with the remainder being 35 or 40 mph. Report on driver yielding rates at multilane cross- ings. Researchers used staged pedestrians and employed the same pro- tocol for crossing and observing driving yield- ing across all sites. -Yielding rates for 22 RRFB sites varied more across dif- ferent cities (34% to 92% yielding) compared with yield- ing rates for TCS (seven sites, 95% to 100% yielding) or PHBs (32 sites, 73 to 92% yielding). -Yielding rates also varied by one- or two-way cross sec- tions, with a decreasing yielding effect for two-way com- pared with one-way, even controlling for total crossing distance. -There was a correlation of increased yielding with higher speed limits, even when two anomalous sites with low volumes and very low driver yielding rates were removed from the analysis. Closer analysis revealed that the results were related to very slight differences in yielding (1 per- centage point between the two averages) on roads with posted speeds of 40 versus 35 mph, and were considered not practically significant. -All of the RRFB sites in this study also had School Crossing signs, which the researchers thought could have contributed to an overall average yielding of 86%, which they indicated was higher than national averages. Fitzpatrick et al. (2014) References for Rectangular Rapid Flash Beacon (RRFB) Domarad, J., P. Grisak, and J. Bolger, “Improving Crosswalk Safety: Rectangular Rapid-Flashing Beacon (RRFB) Trial in Calgary,” In Calgary 2013—The Many Faces of Transportation, Institute of Transportation Engineers, Calgary, AB, Canada, 2013, 10 pp. Fitzpatrick, K., M.A. Brewer, and R. Avelar, “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, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 46–54. Foster, N., C.M. Monsere, and K. Carlos, “Evaluating Driver and Pedestrian Behaviors at Enhanced Multi-lane Midblock Pedestrian Crossings: A Case Study in Portland, OR 4,” Transportation Research Record: Journal of the Transportation Research Board, No. 2464, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 59–66. Hua, J., N. Gutierrez, I. Banerjee, F. Markowitz, and D.R. Ragland, San Francisco PedSafe II Project Outcomes and Lessons Learned, Research Report, Safe Transportation Research & Education Center, eScholarship, University of California, 2009. Hunter, W., R. Srinivasan, and C.A. Martel, Evaluation of the Rectangular Rapid Flashing Beacon at a Pinellas Trail Cross- ing in St. Petersburg, Florida, University of North Carolina Highway Research Center, Chapel Hill, 2009. Pécheux, K., J. Bauer, and P. McLeod, Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reduc- ing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures Final System Impact Report, ITS Joint Program Office, HVH-1, U.S. Department of Transportation, Washington, D.C., 2009, 117 pp. [Online]. Available: http:/safety.fhwa. dot.gov/ped_bike/tools_solve/ped_scdproj/sys_impact_rpt/sys_impact_rpt.pdf. Ross, J., D. Serpico, and R. Lewis, Assessment of Driver Yielding Rates Pre-and Post-RRFB Installation, Bend, Oregon, Report FHWA-OR-RD 12-05, Oregon Department of Transportation, Salem, and Federal Highway Administration, Wash- ington, D.C., 2011, 49 pp. Shurbutt, J. and R. Van Houten, Effects of Yellow Rectangular Rapid-Flashing Beacons on Yielding at Multilane Uncontrolled Crosswalks, Report No. FHWA-HRT-10-043, Federal Highway Administration, McLean, Va., 2010.

135 Van Houten, R., R. Ellis, and E. Marmolejo, “Stutter-Flash Light-Emitting-Diode Beacons to Increase Yielding to Pedestrians at Crosswalks,” Transportation Research Record: Journal of the Transportation Research Board, No. 2073(1), Transportation Research Board of the National Academies, Washington, D.C., 2008, pp. 69–78. Pedestrian Hybrid Beacon PHB Location Type Study Type Safety Effects Citation Pedestrian Hybrid Beacon (PHB) Five PHB sites in urban Tucson, AZ Before–after observational field studies -Red signal or beacon devices (including PHBs, traffic signals, and half signals) were more effective than all other devices evaluated at gaining motorist compliance with formerly uncontrolled crosswalks. -Motorist compliance values ranged between 90% and 100% at all study sites, with an average of 97% across all sites. -Research team found a wide range of motorist compli- ance values relating to the warning devices studied. Fitzpatrick et al. (2006) PHB 21 PHB treatment sites and 102 un-signalized control sites in Tucson, AZ Collision data used for 3 years pretreatment and 3 years following treatment, and for the same periods for the nearby control sites, in order to calculate reduction in expected collisions using the Empirical Bayes method. -A statistically significant reduction in total crashes of 29%, as well as a statistically significant 69% reduction in pedestrian crashes. -A 15% reduction in severe crashes, although the decline was not statistically significant. PHB was effective at leading to a reduction in total col- lisions at treatment sites. Fitzpatrick and Park (2010) PHB Three urban sites in Charlotte, NC Researchers collected observational behavioral effects data from pedestrian crossings during weekday morning and evening peak times at five time points: before the installation, and at 1, 3, 6, and 12 months following installation. -Increase in the percentage of yielding motorists, a decrease in the percentage of trapped pedestrians, and a decrease in pedestrian–vehicle conflicts at all three sites; however, these changes were significant at only one of the three sites. -A statistically significant increase in average vehicle speed was also observed at one of three sites. -Analysis of pre- and post-installation crash data showed no significant change in pedestrian–vehicle crashes, although the sample size was small in both cases. -Changes in pedestrian and motor vehicle actions were more consistent after the PHBs had been in place for 3 months or more. -PHBs were effective at increasing motorist yielding and reducing trapped pedestrians and pedestrian–vehi- cle conflicts. Pulugurtha and Self (2013) 22 sites with RRFBs and school crossing signs The study also exam- ined 32 PHB sites, and 14 sites with TCS. Multilane sites through- out Texas. Focused on 22 sites in three cities, dif- ferent median treatments, numbers of lanes, one- way versus two-way, and posted speed limits from 30 mph to 45 mph. Cross-sectional study of driver yielding rates at treated crosswalks Researchers used staged pedestrians and employed the same protocol for cross- ing and observing driving yielding across all sites. Traffic volumes and pedes- trian volumes were not fac- tored in the analyses. There also appeared to be some confounding between cities, speed limits, and potentially other characteristics. -Over all sites, motorist yielding to PHBs averaged 89%. -Model results indicated that wider crossing distances were associated with slightly higher driver yielding rates at most sites (P < .01). Two-way roads were asso- ciated with lower yielding than one-way (significant at < .10). The posted speed limit did not have a significant predictive effect on driver yielding rates. -Researchers also analyzed the time effect, using a sub- set of the sites from Austin. Results suggested that driver yielding rates increase each day since installa- tion, but these could be associated with unique charac- teristics in Austin. Fitzpatrick, Brewer, and Avelar (2014)

136 References for Pedestrian Hybrid Beacon Fitzpatrick, K., et al., TCRP Report 112/NCHRP Report 562; Improving Pedestrian Safety at Unsignalized Crossings, Trans- portation Research Board of the National Academies, Washington, D.C., 2006, 109 pp. Fitzpatrick, K., M. Brewer, and R. Avelar, “Driving Yielding at Traffic Control Signals, Pedestrian Hybrid Beacons, and Rect- angular Rapid-Flashing Beacons in Texas,” Transportation Research Record: Journal of the Transportation Research Board, No. 2463, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 46–54. Fitzpatrick, K. and E.S. Park, Safety Effectiveness of the HAWK Pedestrian Crossing Treatment, Report FHWAHRT-10-042, Office of Safety Research and Development, Federal Highway Administration, McLean, Va., 2010, 76 pp. Pulugurtha, S.S. and D.R. Self, “Pedestrian and Motorists’ Actions at Pedestrian Hybrid Beacon Sites: Findings from a Pilot Study,” International Journal of Injury Control and Safety Promotion, (ahead-of-print), 2013, pp. 1–10. Install Traffic Signal Without Pedestrian Countdown Timer Traffic Signal/ No PCS Location Type Study Type Safety Effects Citation Five timing patterns: concurrent, none, exclusive, early release, and late release 1,297 urban signalized intersections involving a total of 2,081 pedestrian crashes in 15 U.S. cities A comprehensive study of pedestrian signal heads -Compared with traffic signals without pedestrian sig- nal heads, traffic and pedestrian signals with concur- rent timing had no statistically significant effect on pedestrian crashes. -Exclusive timing was associated with about half of the pedestrian crashes compared with concurrent timing or traffic signals with no pedestrian signals, and this dif- ference was statistically significant. This was only true at locations with pedestrian volumes of more than 1,200 people per day. -There was insufficient sample size with early release or late release signal timing to determine the safety effect of those timing schemes. Mead at al. (2014) cit- ing Zegeer et al. (1982) and Zegeer et al. (1983) Concurrent and exclusive signal tim- ing, compared with no pedestrian signal 320 intersections from three cities in Israel Comparative crash analysis Intersections with exclusive phases for pedestrians had fewer crashes where vehicle and pedestrian volumes were higher. Zaidel and Hocherman (1987) Traffic signals with pedestrian indicators Multiple U.S. cities Literature review and analy- sis of pedestrian crashes -Inconclusive: reduced pedestrian crashes at some locations, had little to no effect at others and may increase crashes at other sites -The presence of a pedestrian signal in itself did not have a statistically significant effect on pedestrian and vehicle delay, but the signal timing scheme did. Robertson and Carter (1984) Installing signals at previously unsignal- ized intersections 447 new traffic signals installed at nonsignalized intersections throughout New York City based on MUTCD warrants Two group pretest posttest research design to compare collision statistics. Pedes- trian collision statistics were collected for the 5 years preceding signal installation and 2 years fol- lowing it, and the authors used ANCOVA in their analysis in order to control for potential regression-to- the-mean effects. Analysis of their results indicated that the average pedestrian crash rate increased by 11.9% at treatment sites and by 60% at comparison sites. This resulted in an ANCOVA-adjusted increase in pedestrian collisions of 1% at signalized sites, results that were not signifi- cant at the 0.05 level. Chen et al. (2013) Traffic signal without pedestrian count- down signal Multiple sites in multiple cities at multilane, mid- block crossings Cross-sectional analysis of behavioral effects compar- ing different signal types -Traffic control signals were associated with high motorist yielding rates when applied at midblock cross- ings. Motorist yielding averaged 98% across seven sites with traffic control signals in three cities. -Yielding rates were lower for pedestrian hybrid bea- cons with an average of 89% motorist yielding across 32 sites in four cities, and for RRFBs at 86% across 22 sites in three cities. Fitzpatrick et al. (2014)

137 References for Install Traffic Signal Without Pedestrian Countdown Timer Chen, L., C. Chen, R. Ewing, C.E. McKnight, R. Srinivasan, and M. Roe, “Safety Countermeasures and Crash Reduction in New York City—Experience and Lessons Learned,” Accident Analysis & Prevention, Vol. 50, 2013, pp. 312–322. Fitzpatrick, K., M. Brewer, and R. Avelar, “Driving Yielding at Traffic Control Signals, Pedestrian Hybrid Beacons, and Rect- angular Rapid-Flashing Beacons in Texas,” Transportation Research Record: Journal of the Transportation Research Board, No. 2463, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 46–54. Mead, J., C. Zegeer, and M. Bushell, “Evaluation of Pedestrian-Related Roadway Measures: A Summary of Available Research,” Pedestrian and Bicycle Information Center, Chapel Hill, N.C., 2014 [Online]. Available: http:/www.pedbikeinfo. org/cms/downloads/PedestrianLitReview_April2014.pdf. Robertson, H.D. and E.C. Carter. The Safety, Operational, and Cost Impacts of Pedestrian Indications at Signalized Intersec- tions,” Transportation Research Record: Journal of the Transportation Research Board, No. 959, Transportation Research Board, National Research Council, Washington, D.C., 1984. Zaidel, D.M. and I. Hocherman, “Safety of Pedestrian Crossings at Signalized Intersections,” Transportation Research Record: Journal of the Transportation Research Board, No. 1141, Transportation Research Board, National Research Coun- cil, Washington, D.C., 1987, pp. 1–6. Install Traffic Signal with Pedestrian Countdown Signal (PCS) Traffic Signal with PCS Location Type Study Type Safety Effects Citation The first comparison group, about half (629) of the control intersections that were planned to receive PCS, but had not yet received the treatment. The sec- ond comparison group (628 intersec- tions) were not scheduled to receive PCS. Treatment: 14 pilot loca- tions with PCS in San Francisco, CA Comparison group 1: 629 intersections Comparison group 2: 628 intersections Subset of treatment group: seven intersec- tions with 2+ crashes Subset of comparison group: 185 intersections with 2+ crashes Before–after: computed the ratio of pedestrian injury events for 21 months fol- lowing installation of PCS to the number that occurred 21 months before installa- tion and for several compar- ison groups -The number of pedestrian injury collisions declined by a statistically significant amount following the intro- duction of PCS (48% ratio of after to before injuries), whereas no significant declines were found in the two comparison groups (97% and 90%). -Subset of treatment group ratio: 42.3% -Subset of comparison group ratio: 55.6% -An additional comparison of the two high-crash sub- sets revealed that a significant part of the effect was likely to the result of regression toward the mean (RTM), since a significant (but smaller) decline was also observed in the high-crash group that had not been treated (56% ratio of before to after pretreatment inju- ries) compared with the high-crash group that received the PCS (42% ratio of after to before injuries). The treated intersections were selected on the basis of pre- treatment injury levels, which could contribute toward an RTM effect. -The decrease in pedestrian crashes was greater for the treated intersections with high crashes compared with the untreated group with high crashes. Markowitz et al. (2006) Traffic signal with pedestrian count- down signal The researchers com- pared the rate of pedes- trian–motor vehicle colli- sions at 1,965 Toronto signalized intersections before and after the installations. Population-based study to evaluate the impact of wide- spread replacement of tradi- tional signals with PCS on pedestrian–motor vehicle collisions. A total of 9,262 pedestrian–vehicle colli- sions took place during the 10-year study period. -Initial study results indicated that installation of PCS had no statistically significant effect on the number of pedestrian–motor vehicle collisions at the intersections where they were installed. Camden et al. (2011)

138 Traffic Signal with PCS Location Type Study Type Safety Effects Citation Traffic signal with pedestrian count- down signal Reanalysis of Camden et al. (2011) study; same 1,965 Toronto signalized intersections Reanalysis of Camden et al. data -A baseline pedestrian–motor vehicle crash rate for each intersection (which was required to be treated with a traditional, non-PCS signal before the PCS installation for inclusion) was used for its own control, along with time and season covariates. -PCS were associated with a 26% increase in pedes- trian–motor vehicle collisions compared with tradi- tional signals (incidence rate ratio = 1.26, 95% CI, 1.11 to 1.42). Richmond et al. (2014) Traffic signal with pedestrian count- down signal Used monthly crash data collected from 2001– 2010 to compare the effect of introducing PCS to 362 signalized inter- sections, with 87 control sites, in Detroit Crash-based, citywide study was conducted to assess effects on total crashes. Authors suggested that about 17% total crash reduc- tions were attributable to the implementation of PCS, but these results need validation. Huitema et al. (2014) Traffic signal with pedestrian count- down signal Assessed pedestrian safety effects of wide- spread installation of new traffic signals at formerly unsignalized New York City intersections accord- ing to MUTCD (2009) warrants. A total of 447 intersections were in the treatment group, and 442 in the comparable refer- ence (untreated) group. 5 years of before period crashes and 2 years of after period crashes were analyzed. Negative binomial models with General Estimating Equations method controlled for time trends and a variety of other intersection and environmental factors. In addition to using nearby ref- erence locations, the authors mention that although they could not control for traffic volumes, volumes through- out the city in general have remained relatively stable, and at capacity, for 10 years. Built environment and site factors thought to be related to safety outcomes were controlled through the analy- sis models. -Pedestrian crashes increased at both treated and com- parison intersections, but by a smaller amount at treated (+12%) than comparison (unsignalized) (+67%) locations. These observed data resulted in model esti- mates of effectiveness of -47% fewer pedestrian–vehicle collisions for installing signals compared with not installing signals. Observed declines for motor vehicle–motor vehicle collisions resulted in model estimates of -60% fewer vehicle–vehicle collisions for installing signals compared with not installing signals. Both of these effects were significant. Chen et al. (2014) Split-phase (pro- tected pedestrian crossing and pro- tected left-turn phases) timing 30 treatment intersec- tions; 493 untreated com- parison intersections in New York City 5 years of before period crashes and 2 years of after period crashes were analyzed. Same methods as Chen, Chen, and Ewing (2014) for installing traffic signals Reductions in observed pedestrian crashes of 39% at treatment sites and 8% at comparison sites resulted in an adjusted estimate of the effects of split-phase timing of -47% reduction in pedestrian crashes due to the treat- ment, and -30% reduction in multiple motor vehicle crashes. Both of these effects were significant. Chen et al. (2014) Barnes Dance (all walk phase) 36 treatment intersec- tions; 516 untreated com- parison intersections in New York City 5 years of before period crashes and 2 years of after period crashes were analyzed. Same methods as Chen, Chen, and Ewing (2014) for installing traffic signals Average pedestrian crashes decreased by a larger degree, 51% at Barnes Dance (pedestrian scramble, or All Walk signal with crossing allowed in any direction including diagonally) sites versus 9% reduction at comparison sites. These measures yielded a model- adjusted estimate of effect (using the same methods as above) of: -55% reduction in pedestrian crashes compared with not implementing the Barnes Dance. This effect was significant. -14% higher motor vehicle–motor vehicle crashes relating to Barnes Dance implementation, but the effect was not significant. Chen et al. (2014)

139 Traffic Signal with PCS Location Type Study Type Safety Effects Citation Increased pedestrian crossing time Implemented at 244 intersections in New York City; 1,173 untreated comparison intersections 5 years of before period crashes and 2 years of after period crashes were analyzed. Same methods as Chen, Chen, and Ewing (2014) for installing traffic signals Where pedestrian crossing time was increased, the average pedestrian crash rate decreased by 50%, whereas at the control sites it decreased by 4%. These observed differences resulted in model predicted effects of: -65% reduction in pedestrian crashes compared with not increasing the pedestrian cycle length. -11% reduction in multiple motor vehicle collisions also resulted, but this downward trend was not statisti- cally significant. Chen et al. (2014) Install traffic signal with pedestrian countdown signal 447 formerly unsignal- ized New York City intersections according to MUTCD (2009) war- rants. A total of 447 intersections were in the treatment group, and 447 in the comparable refer- ence (untreated) group. 5 years of before period crashes and 2 years of after period crashes were ana- lyzed with comparison group. Regression toward the mean and other trend affects were accounted for by the ANCOVA pretest posttest and comparison group models. These models attempted to control for a variety of other site characteristics and built environment factors by matching these for compari- son groups. In addition to using nearby reference loca- tions, the authors mention that although they could not control for traffic volumes, volumes throughout the city in general remained rela- tively stable, and at capac- ity, for 10 years. -Pedestrian crashes increased at both treated and com- parison intersections, but by a smaller amount at treated (+12%) than comparison (unsignalized) (+60%) locations. -The observed data resulted in a slight predicted increase in pedestrian crashes (ANCOVA model esti- mated CMF of 1.01) for installing signals compared with not installing signals. The effect was not statisti- cally significant. -Vehicle–vehicle crashes declined at both treated and untreated intersections: by 49% at treated and 14% at untreated intersections. -The observed data resulted in a CMF estimate of 0.78 (estimated 22% reduction) for vehicle–vehicle colli- sions for installing signals (effect was statistically significant). Chen et al. (2013) (ear- lier study of effects of signals and signal tim- ing strategies) Split-phase timing (protected left-turn and protected pedes- trian phases) 30 intersections in treat- ment group; 579 in untreated comparison group 5 years of before period crashes and 2 years of after period crashes were ana- lyzed, with untreated com- parison group. Same methods as Chen et al. (2013) for installing traffic signals -Average observed pedestrian crashes decreased by 39% at treated locations and by 12% at the untreated locations. -The effects of split-phase timing were an estimated CMF of 0.74 for pedestrian crashes due to the treat- ment; however, the effect was not statistically significant. -There was also a downward but not statistically signif- icant reduction in multiple motor vehicle crashes. Chen et al. (2013) (ear- lier study of effects of signals and signal tim- ing strategies) Barnes Dance (all walk phase) 37 intersections in treat- ment group; 4,266 in untreated comparison group 5 years of before period crashes and 2 years of after period crashes were ana- lyzed, with untreated com- parison group. Same methods as Chen et al. (2013) for installing traf- fic signals -Average observed pedestrian crashes decreased by 44% at the treated locations and by 9% at the untreated locations. -The model-predicted CMF estimate was a significant 0.65 compared with not implementing the Barnes Dance. -There was no significant effect on motor vehicle– motor vehicle crashes. Chen et al. (2013) Increase pedestrian crossing time 244 intersections in treat- ment group; 915 intersec- tions in untreated com- parison group 5 years of before period crashes and 2 years of after period crashes were ana- lyzed, with untreated com- parison group. Same methods as Chen et al. (2013) for installing traf- fic signals. -Where pedestrian crossing time was increased, the average observed pedestrian crash rate decreased by 50%, whereas at the control sites it decreased by 29%. -These observed differences resulted in a model pre- dicted CMF of 0.49 for pedestrian crashes compared with not increasing the pedestrian cycle length. -A slight increase in multivehicle crashes was not sta- tistically significant. Chen et al. (2013)

140 Traffic Signal with PCS Location Type Study Type Safety Effects Citation Impact of PCS (pedestrian count- down timers) Eight large intersections in south Florida (19,000– 65,000 vehicles per day) Before–after pedes- trian behavior com- parisons, unknown period -Increase in the percentage of pedestrians complying with the “Walk” indication (entered during the “Walk” phase and not during flashing or steady “Don’t Walk”) by a slight, but nonsig- nificant amount (from 55% to 56.3%) -Significant decrease (from 13.7 to 8.1%) in pedestrians enter- ing during the flashing “Don’t Walk” -These differences were apparently offset by a larger increase in pedestrians entering during the steady “Don’t Walk” phase (noncompliant) (from 31.3 to 35.5%; attributed to three locations). -The percentage of successful crossings (completed before steady “Don’t Walk), all intersections combined, increased sig- nificantly, from 56.15% to 63.27%. -Data unavailable to assess potential effects on driver behaviors. Reddy et al. (2008) Traffic signal with pedestrian count- down signal 20 crosswalks at five intersections in Mont- gomery County, MD Before–after study of behavioral effects and a survey -Survey revealed that most pedestrians were aware of the countdown signal and a majority (63%) also understood its meaning. -At six of the 20 crosswalks, there were statistically significant increases in pedestrians entering the crosswalk during the “Walk” phase; in two of the 20 crosswalks observed, there was a statistically significant decrease in the number of pedestrians who entered on the “Walk” indication. -Three of the five intersections experienced a statistically sig- nificant decrease in the number of phases with pedestrians left in the crosswalk at the release of conflicting traffic. Although the other two intersections experienced an increase, the increases were slight and were not statistically significant. -Number of conflicts between pedestrians and vehicles decreased (statistically significant) in the after periods at four of five intersections observed. -No effect on nonplatooning motorist approach speeds on a late green Eccles et al. (2004) PCS with the addi- tion of animated eyes. One high-volume arterial intersection in Las Vegas with a significant crash history. Crash issues included drivers failing to yield, nighttime crashes, and a high percentage of elderly pedestrians. Before–after study of pedestrian behavior -Percentage of pedestrians beginning their crossing during the “Walk” phase increased significantly. -Significant increase in the percentage of pedestrians who looked for vehicles before beginning to cross -Significant increase in those beginning to cross during the “Walk” phase -Percentage of signal cycles in which the call button was pushed also increased significantly. -Not possible to ascertain how much of the effects related to the PCS and how much to the animated eyes. Mead et al. (2014); Vasudevan et al. (2011) Traffic signal with pedestrian count- down signal Urban. Lake Buena Vista, FL. Two intersec- tions with countdown signals and three control intersections without countdown signals. Traf- fic volume ranged from 30,000 to 68,000. Cross-sectional comparative study of sites with PCS versus without -Pedestrians were less likely to comply with the walk signal at the countdown signal locations on multilane, high-ADT sites (speed limits 35 or 50 mph) compared with similar non-PCS sites, and the difference was statistically significant. -At one multilane comparison site, a slightly smaller percentage of pedestrians ran out of time and remained in the crossing at the signal change than at the PCS locations, but the difference was not statistically significant. -3% of pedestrians began running at the flashing “Don’t Walk” at the PCS locations compared with 10% at the control loca- tions, and the difference was statistically significant. Huang and Zegeer (2000) Traffic signal with pedestrian count- down signal Four intersections along an arterial corridor in Lawrence, KS Continuous speed data were collected on approaching traffic at two inter- sections with PCS and two without to study driver deci- sions in response to PCS. -Drivers were significantly less likely to increase their speed in order to reach the intersection before the beginning of the red phase when PCS were present, and some drivers began to slow to a stop before the beginning of the amber phase when PCS were present. Some drivers make safer driving decisions with PCS, even though the countdown information was not intended to be used by them. Schrock and Bundy (2008)

141 References for Install Traffic Signal with Pedestrian Countdown Signal (PCS) Camden, A., R. Buliung, L. Rothman, C. Macarthur, and A. Howard, “The Impact of Pedestrian Countdown Signals on Pedestrian–Motor Vehicle Collisions: A Quasi-Experimental Study,” Injury Prevention: Journal of the International Society for Child and Adolescent Injury Prevention, Vol. 18, No. 4, 2011, pp. 210–215 [Open Access]: http:/injuryprevention.bmj.com/ content/18/4/240. Chen, L., C. Chen, and R. Ewing, “The Relative Effectiveness of Signal Related Pedestrian Countermeasures at Urban Inter- sections—Lessons from a New York City Case Study,” Transport Policy, Vol. 32, 2014, pp. 69–78. Chen, L., C. Chen, R. Ewing, C.E. McKnight, R. Srinivasan, and M. Row, “Safety Countermeasures and Crash Reduction in New York City—Experience and Lessons Learned,” Accident Analysis & Prevention, Vol. 50, 2013, pp. 312–322. Eccles, K., R. Tao, and B. Mangum, “Evaluation of Pedestrian Countdown Signals in Montgomery County, Maryland,” Trans- portation Research Record: Journal of the Transportation Research Board, No. 1878, Transportation Research Board of the National Academies, Washington, D.C., 2004, pp. 36–41. Huang, H. and C. Zegeer, The Effects of Pedestrian Countdown Signals in Lake Buena Vista, Florida Department of Trans- portation, Tallahassee, 2000. Huitema, B., R.G. Van Houten, and H. Manal, “An Analysis of the Effects of Installing Pedestrian Countdown Timers on the Incidence of Pedestrian Crashes in the City of Detroit, Michigan,” Transportation Research Board 93rd Annual Meeting Compendium of Papers, Paper No. 14-0227, 2014. Markowitz, F., S. Sciortino, J. Fleck, and B. Yee, “Pedestrian Countdown Signals: Experience with an Extensive Pilot Instal- lation,” ITE Journal, Vol. 76, No. 1, 2006, pp. 43–47. Mead, J., C. Zegeer, and M. Bushell, “Evaluation of Pedestrian-Related Roadway Measures: A Summary of Available Research,” Pedestrian and Bicycle Information Center, Chapel Hill, N.C., 2014 [Online]. Available: http:/www.pedbikeinfo. org/cms/downloads/PedestrianLitReview_April2014.pdf. Reddy, V., T. Datta, D. McAvoy, P. Savolainen, M. Abdel-Aty, and S. Pinapaka, Evaluation of Innovative Safety Treatments, Volume 5: A Study of the Effectiveness of Countdown Pedestrian Signals, Florida Department of Transportation, Tallahassee, 2008 [Online]. Available: http:/www.dot.state.fl.us/research-center/completed_proj/summary_sf/bd500/bd500_v5_rpt.pdf [accessed May 2015]. Richmond, S.A., A.R. Willan, L. Rothman, A. Camden, R. Buliung, C. Macarthur, and A. Howard, “The Impact of Pedestrian Countdown Signals on Pedestrian–motor Vehicle Collisions: A Reanalysis of Data from a Quasi-experimental Study,” Injury Prevention, Vol. 20, No. 3, 2014, pp. 155–158 [Open Access]: http:/injuryprevention.bmj.com/content/20/3/155.full. Schrock, S.D. and B. Bundy, “Pedestrian Countdown Timers: Do Drivers Use Them to Increase Safety or to Increase Risk Taking?” Transportation Research Board 87th Annual Meeting Compendium of Papers, Paper No. 08-2203, 2008. Vasudevan, V., S.S. Pulugurtha, S.S. Nambisan, and and M.R. Dangeti, “Effectiveness of Signal-Based Countermeasures for Pedestrian Safety,: Findings from a Pilot Study,” Transportation Research Record: Journal of the Transportation Research Board, No. 2264, Transportation Research Board of the National Academies, Washington, D.C., 2011, pp. 44–53.

142 Leading Pedestrian Interval (LPI) LPI Location Type Study Type Safety Effects Citation Leading pedestrian interval Multiple sites in New York City Evaluation of the LPI on pedestrian crash statistics. Data collected for the 5 years preced- ing and 5 years fol- lowing LPI instal- lation for treatment and control intersections. -LPI had the effect of decreasing both collision occurrence and severity at treatment intersections, especially at intersections with heavy turning volumes. -Results of this study and others led the New York City Depart- ment of Transportation to decide to install more LPIs as a pedestrian safety countermeasure. King (1999) Leading pedestrian interval 10 intersections in State College, PA. 14 compari- son intersections similar in terms of site character- istics, traffic and pedes- trian volumes, and crash data. Before–after crash analysis Reduction of 59% in pedestrian–vehicle crashes at the 10 treated intersections (significant at the 95% confidence level) Fayish and Gross (2010) Leading pedestrian interval Urban San Francisco high-crash areas. Four intersections with LPIs. Before–after study (B: 1–4 years/A: 1–2 years) of behavioral effects Significant reduction in the percentages of vehicles turning in front of pedestrians, from 6.2% pretreatment to 4% posttreat- ment on average over four sites with 4-s pedestrian head starts Hua et al. (2009) Leading pedestrian interval Suburban, high-volume intersection in Anaheim, CA Comparison study of “compromised” crossings with and without LPI of 3 s at same crosswalk, data collected December 2006– January 2007 Compromise was defined as one in which the pedes- trian is delayed by a turning vehicle or changes travel path or speed in response, and was measured for pedestrians waiting on the curb and in the crosswalk. -The proportion of pedestrians with crossings compromised while still waiting on curb at beginning of crossing (before they could step out, had “Walk” signal, and motorist still had red) was significantly higher with LPI, for all three volume condi- tions: low right-turn volumes (1–5 vehicles in queue), high right-turn volumes (6+ in queue), and all right-turn volumes. -The proportion of “Walk” cycles with pedestrians compro- mised while still waiting on curb increased significantly for all three volume conditions. -The proportion of pedestrians with crossings compromised while in crosswalk was significantly lower for high and all vol- umes, but significantly higher for low-volume conditions with LPI in place. -The proportion of “Walk” cycles with pedestrians compro- mised while in crosswalk also was lower for high and all vol- ume conditions, but (nonsignificantly) higher for lower-volume conditions. Hubbard et al. (2008) Leading pedestrian interval Three signalized intersec- tions in St. Petersburg, FL, with a 1-s all-red interval Field observations Reduced conflicts and reduced percentage of pedestrians yield- ing to turning motor vehicles for 3-s head start. Van Houten et al. (2000) Leading pedestrian interval Two intersections in urban Miami, FL Before–after study (time periods unknown) Increase in the percentage of yielding by left-turning drivers; a statistically significant increase in the percentage of pedestrians who pushed the call button and who crossed during the first 4 s of the walk phase Pecheux et al. (2009) References for Leading Pedestrian Interval (LPI) Fayish, A.C. and F. Gross, “Safety Effectiveness of Leading Pedestrian Intervals Evaluated by a Before-After Study with Comparison Group,” Transportation Research Record: Journal of the Transportation Research Board, No. 2198, Transporta- tion Research Board of the National Academies, Washington, D.C., 2010, pp. 15–22. Hua, J., N. Gutierrez, I. Banerjee, F. Markowitz, and D.R. Ragland, San Francisco PedSafe II Project Outcomes and Lessons Learned, Research Report, Safe Transportation Research & Education Center, eScholarship, University of California, 2009. Hubbard, S.M.L., D.M. Bullock, and J.H. Thai, “Trial Implementation of a Leading Pedestrian Interval: Lessons Learned,” ITE Journal, Vol. 78, No. 10, pp. 32–41.

143 King, M.R., “Calming New York City Intersections,” Transportation Research Circular E-C019: Urban Street Symposium Conference Proceedings, Dallas, Tex., June 28–30, 1999, I-3, pp. 1–15. Pécheux, K., J. Bauer, and P. McLeod, Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reduc- ing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures Final System Impact Report, ITS Joint Program Office, HVH-1, U.S. Department of Transportation, Washington, D.C., 2009, 117 pp. [Online]. Available: http:/safety.fhwa. dot.gov/ped_bike/tools_solve/ped_scdproj/sys_impact_rpt/sys_impact_rpt.pdf. Van Houten, R.V., R.A. Retting, C.M. Farmer, and J. Van Houten, “Field Evaluation of a Leading Pedestrian Interval Signal Phase at Three Urban Intersections,” Transportation Research Record: Journal of the Transportation Research Board, No. 1734, Transportation Research Board, National Research Council, Washington, D.C., 2000, pp. 86–92. “Right Turn on Red” (RTOR) Restrictions NTOR Location Type Study Type Safety Effects Citation Western RTOR Urban and suburban New York, Ohio, and Wisconsin; City of New Orleans, LA; Los Angeles, CA. Seven datasets of crash data at numer- ous sites. Comparison of computerized acci- dent data from the periods before and after implementa- tion of the RTOR rule. Up to 4.5 years in after period. Before period was as much crash data as was available. In all cases at least 12 months. A second part of this study involved analysis of police crash reports. The following increases in crash rates after the RTOR rule were observed: -43% for pedestrian accidents and 82% for bicycles in New York State; 107% for pedestrians and 72% for bicycles in Wisconsin; 57% for pedestrians and 80% for bicycles in Ohio; 82% for pedes- trians in New Orleans. -Given the rarity of RTOR–pedestrian crashes, these percentages represented increases in very small numbers. -Authors were able to identify a common crash scenario involving RTOR: Often, a driver who is stopped before turning right focuses on traffic coming from the left in order to identify a gap adequate to permit his right turn. Consequently, the motorist does not see a pedestrian or bicyclist on his right and a conflict occurs when the turn is initiated. Preusser et al. (1982) “No Turn on Red” restrictions 110 intersections in Washington, DC; Dal- las and Austin, TX; Detroit, Lansing, and Grand Rapids, MI Collected observa- tional data on more than 67,000 drivers, looking for links between motorist violations of “No Turn on Red” (NTOR) signs and the related crashes with pedestrians. -3.7% of all drivers making a right turn violated the NTOR signs. When given an opportunity to violate the NTOR sign (i.e., being the first car in line at a signalized intersection with no pedestrians in front of them and no cars coming from the left), 21% of the drivers ignored the NTOR signs. -23.4% of all RTOR violations create a conflict with a pedestrian. Where RTOR is permitted, 56.9% of drivers do not come to a complete stop before turning, compared with 68.2% who fail to do so at STOP-controlled intersections. Zegeer and Cynecki (1986) “No Turn on Red” restrictions Seven of the counter- measures for intersec- tions with RTOR were tested at 34 intersec- tions in six cities in the U.S. In a later phase of the study, Zegeer and Cynecki devel- oped 30 potential countermeasures to enhance pedestrian safety at intersec- tions permitting RTOR. -NTOR signs with a red ball were more effective than standard black-and-white ones. -An offset stop bar increased compliance in making a full stop before turning at RTOR locations and also lessened conflicts with traffic on cross streets. -The more costly electronic NTOR/black-out sign used only dur- ing school crossing periods or other critical times was slightly more effective than the regular NTOR sign. -Drivers were more likely to comply with the RTOR restriction if it was limited to peak pedestrian periods rather than imposed full time. -In areas with moderate or low RTOR volumes, an alternative “No Turn on Red When Pedestrians Are Present” sign was effective at intersections with low to moderate volumes of RTOR vehicles. In general, the likelihood of a RTOG accident was found to be greater than that of a RTOR accident, based on conflict data. Zegeer and Cynecki (1986)

144 NTOR Location Type Study Type Safety Effects Citation “No Turn on Red” Restrictions 26 sites in the Province of Quebec where RTOR was authorized for a period of 9 months Two-part pilot study: Part 1—observing driver behavior; Part 2—collect- ing data from a number of U.S. and Canadian agen- cies concerning the effect that RTOR had on safety and on traffic operations -In most cases, RTOR does not pose a danger to motorists, cyclists, or pedestrians. Lord reported that pedestrian crashes involving a RTOR maneu- ver make up less than 1% of all reported accidents in the United States and Canada, and the crashes that do occur are usually not severe. Many of the transportation experts and researchers who were surveyed for this study did not consider RTOR to be a safety problem. -It was less clear that RTOR does not contribute to conflicts, increase delay to pedestrians, and lower the quality of service at signalized intersections Mead et al. (2014) cit- ing Lord (2003) Restrictions on right turn on red Multiple intersections in Ala- bama and South Carolina Before–after crash data analysis -Statistically significantly increase during the after period in South Carolina for right-turning property damage accidents than for accidents not involving right turns, but not in Alabama -No statistically significant difference in the rate of change in fatality or injury accidents in either state when comparing right-turning vehicles to non-right- turning vehicles. Furthermore, there was no evi- dence of increased pedestrian accidents resulting from RTOR in either South Carolina or Alabama. Clark et al. (1983) Traffic signs that limit RTOR during specific time periods and highly visible traffic signs that pro- hibit RTOR when pedestrians are present 15 signalized intersections in in Arlington County, VA. A third of the intersections (five sites) served as the control group, while the others were equally divided between the two treatments. Before–after study of behavioral effects compar- ing sites. At the first group of five treated sites, signs were placed that stated “No Turn on Red, 7 a.m.–7 p.m., Mon–Fri.” At the second group of five treated sites, fluorescent yellow-green reflective signs reading “No Turn on Red – When Pedestrians Are Present” were implemented. -During the pretreatment period, 80% of all observed vehicles turned right on red at these loca- tions. During the before period, 39% of all vehi- cles observed did not come to a full stop before making a RTOR, and 14% of pedestrians yielded to motorists who were making RTORs. -Following sign installation, there was a small decline in the percentage of motorists who turned right on red at the sites with signs restricting right turns when pedestrians are present, and a large decrease where RTOR was not permitted during specified time periods. -In terms of pedestrian behavior, after installation of the signs, there was a large decrease in pedestri- ans who yielded to vehicles turning right at time- specific locations, with little change at sites that prohibited RTOR in the presence of pedestrians. -Overall, the researchers concluded that signs that made RTOR dependent on driver discretion related to the presence or absence of pedestrians were less effective. Retting et al. (2002) Three types of “No Turn on Red” (NTOR) signs One site in Miami, FL, tested with three signs. The first two that were used were static signs printed with the messages “No Turn on Red” and “No Turn on Red When Pedestrians in Crosswalk.” The third sign was an electric sign that was dark during the protected right-turn phase, displayed the message “Yield to Pedestrians” during the green phase, and displayed a “No Turn on Red” message during phases when a right turn was prohibited and pedestrians had pushed the call button. Data were collected in four phases. Baseline data were collected with the condi- tional “No Turn on Red When Pedestrians in Crosswalk” in place. For the next phase, the condi- tional sign was replaced with the “No Turn on Red” sign. For the second phase the active sign was used, and for the third phase the baseline conditional sign was reinstalled. Violations were lowest when the active sign was used. Of those drivers who violated the turn restriction, there was a significant increase in driv- ers coming to a full stop before doing so from 29% during the baseline phase (with the static, condi- tional sign) to 78% with the electronic sign. The authors concluded that the electronic NTOR sign was of medium effectiveness in increasing pedes- trian safety, whereas the two static signs were of low effectiveness. Pecheux et al. (2009) Restrictions on right turn on red The sign was installed concur- rently with a high-visibility crosswalk at one intersection in Las Vegas, NV. Before–after study of behavioral effects No significant improvement in the percentage of drivers who yielded to pedestrians, but a signifi- cant increase in the percentage of drivers who came to a complete stop before making a RTOR, and in the percentage of pedestrians who looked for a vehicle before crossing. The authors con- cluded that the “Intelligent Transportation System No Turn on Red” sign made drivers and pedestri- ans more cautious at the intersection. Dangeti et al. (2010)

145 References for “Right Turn on Red” (RTOR) Restrictions Clark, J.E., S. Maghsoodloo, and D.B. Brown, “Public Good Relative to Right-Turn-on-Red in South Carolina and Alabama,” Transportation Research Record 926, Transportation Research Board, National Research Council, Washington, D.C., 1983, pp. 24–31. Dangeti, M.R., S.S. Pulugurtha, V. Vasudevan, S.S. Nambisan, and O.C. White, Jr., “Evaluating ITS-Based Countermeasures: How Effective Are They in Enhancing Pedestrian Safety?” Transportation Research Board 89th Annual Meeting Compen- dium of Papers DVD, Paper No. 10-2217, Washington, D.C., 2010. Lord, D., “Synthesis on the Safety of Right Turn on Red in the United States and Canada,” presented at 82nd Annual Meeting of the Transportation Research Board, Washington, D.C., 2002. Mead, J., C. Zegeer, and M. Bushell, “Evaluation of Pedestrian-Related Roadway Measures: A Summary of Available Research,” Pedestrian and Bicycle Information Center, Chapel Hill, N.C., 2014 [Online]. Available: http:/www.pedbikeinfo. org/cms/downloads/PedestrianLitReview_April2014.pdf. Pécheux, K., J. Bauer, and P. McLeod, Pedestrian Safety Engineering and ITS-Based Countermeasures Program for Reduc- ing Pedestrian Fatalities, Injury Conflicts, and Other Surrogate Measures Final System Impact Report, ITS Joint Program Office, HVH-1, U.S. Department of Transportation, Washington, D.C., 2009, 117 pp. [Online]. Available: http:/safety.fhwa. dot.gov/ped_bike/tools_solve/ped_scdproj/sys_impact_rpt/sys_impact_rpt.pdf. Preusser, D.F., W.A. Leaf, K.B. DeBartolo, R.D. Blomberg, and M.M. Levy, “The Effect of Right-Turn-on-Red on Pedestrian and Bicyclist Accidents,” Journal of Safety Research, Vol. 13, No. 2, 1982, pp. 45–55. Retting, R.A., M.S. Nitzburg, C.M. Farmer, and R.L. Knoblauch,“Field Evaluation of Two Methods for Restricting Right Turn on Red to Promote Pedestrian Safety,” ITE Journal, Vol. 72, No. 1, 2002, pp. 32–36. Zegeer, C. V. and M. J. Cynecki, “Evaluation of Countermeasures Related to RTOR Accidents that Involve Pedestrians.” In Transportation Research Record: Journal of the Transportation Research Board, No. 1059, Transportation Research Board of the National Academies, Washington, D.C., 1986, pp. 24–34.

Abbreviations and acronyms used without definitions in TRB publications: A4A Airlines for America AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACI–NA Airports Council International–North America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FAST Fixing America’s Surface Transportation Act (2015) FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers MAP-21 Moving Ahead for Progress in the 21st Century Act (2012) NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TDC Transit Development Corporation TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation

TRANSPORTATION RESEARCH BOARD 500 F ifth S treet, N .W . W ashing to n, D .C . 20001 A D D R ESS SER VICE R EQ UESTED