EVALUATION OF THE EFFECTIVENESS OF PEDESTRIAN COUNTDOWN SIGNALS

Transcription

1 EVALUATION OF THE EFFECTIVENESS OF PEDESTRIAN COUNTDOWN SIGNALS By DEBORAH LINDORO LEISTNER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2005

2 Copyright 2005 by Deborah Lindoro Leistner

3 This document is dedicated to my father.

4 ACKNOWLEDGMENTS I thank my fiancé, Douglas Robinson, for his patient love and care, and for all of his support during the completion of my degree over the past two years. I thank Dr. Scott Washburn for the thoughtful direction and guidance, and the members of my committee, Dr. Lily Elefteriadou and Dr. Ruth Steiner, for participating in this project and sharing their knowledge and expertise. I thank Brian Kanely, Transportation Services Manager at the City of Gainesville Public Works Department, for allowing me the necessary flexibility to complete this project. I also thank Phil Mann, Conrad Renshaw, and Kris McCoy. Finally, I thank the University of Florida students that contributed in many ways to this effort: Byungkon Ko, Christian Gyle, Jessica Morriss, and Jennifer Webster iv

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS... iv LIST OF TABLES... vii LIST OF FIGURES... viii ABSTRACT... ix CHAPTER 1 INTRODUCTION LITERATURE REVIEW...5 Meaning of Pedestrian Signal Indications...5 Understanding of Pedestrian Signal Indications...6 Conventional vs. Countdown Pedestrian Signal Indications...7 Influence on Pedestrian Behavior RESEARCH APPROACH...12 Site Selection...12 Site Description...13 E University Avenue and E 1 st Street...13 W University Avenue and W 2 nd Street...14 W University Avenue and W 17 th Street...14 W University and North-South Drive...15 Archer Road and SW 16 th Street...15 Data Collection...16 Data Reduction RESULTS AND ANALYSIS...23 Performance Measures...23 Statistical Analysis...24 Discussion of Results by Performance Measure...25 Pedestrians Entering the Crosswalk...25 Pedestrians Exiting the Crosswalk...29 v

6 Compliance with FDW Indication...30 Erratic Pedestrian Behavior...31 Pedestrian-Vehicle Conflicts CONCLUSIONS AND RECOMMENDATIONS...37 Summary of Conclusions...39 Recommendations for Future Study...39 APPENDIX A DATA COLLECTION SAMPLE...41 B SAMPLE PROCESSED DATA FILE...65 LIST OF REFERENCES...71 BIOGRAPHICAL SKETCH...73 vi

7 LIST OF TABLES Table page 3-1 Data Collection Periods Used for Observation and Analysis Calculated Test Statistic by Performance Measure Pedestrians Entering Crosswalk Pedestrians Exiting Crosswalk Compliance with the Flashing Don't Walk Indication Erratic Behavior During Crossing Pedestrian-Vehicle Conflicts...34 vii

8 LIST OF FIGURES Figure page 1-1 Conventional Pedestrian Signal Indications Pedestrian Countdown Signal Indications Schematic of Count Station Locations Signage for Pedestrian Crossings Intersections in the Downtown Area Intersections Adjacent to UF campus Intersection of Archer Rd & SW 16 th St Location of Camera Installation and Field of View at Each Study Intersection Composite Image Capture...19 viii

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science EVALUATION OF THE EFFECTIVENESS OF PEDESTRIAN COUNTDOWN SIGNALS By Deborah Lindoro Leistner August 2005 Chair: Scott S. Washburn Major Department: Civil and Coastal Engineering According to the National Highway Safety Administration, thousands of pedestrians are injured or killed in collisions with vehicles every year. Approximately one quarter of all fatalities occur at urban intersections, and the main cause of fatalities is improper crossing. Pedestrians misunderstanding of pedestrian signal indications at crossings is identified in the literature as a contributing factor to improper crossings. To address this problem and potentially increase the safety at signalized intersection crossings, pedestrian countdown signals were created by incorporating a countdown timer that is displayed simultaneously during the flashing DON T WALK (FDW) interval. The timer counts down the number of seconds remaining until the display of the solid DON T WALK (DW) interval. It has been purported that this signal design leads to a higher level of pedestrian safety by enabling pedestrians to make better crossing decisions with the added information. ix

10 A before-and-after study of pedestrian countdown signals was conducted at five intersections in Gainesville, Florida. All of the intersections had high pedestrian and vehicular traffic volumes during certain times of the day. Additionally, several distinct pedestrian populations were present across the intersections. The data were collected from October 2003 to April 2004, using a video detection system that simultaneously captures pedestrian and vehicular movements and the corresponding traffic signal indications. Over the course of the study, a total of 7,639 pedestrians were observed before and 6,339 pedestrians were observed after the installation of the countdown signals. Overall, the countdown signals appear to have had a positive influence on pedestrian crossing behavior. At most of the study intersections, the proportion of pedestrians entering the crosswalk on the WALK indication increased, while the proportion entering on the steady DW indication decreased. The countdown signals have also had the effect of increasing the proportion of pedestrians exiting on the FDW interval as opposed to the DW interval, therefore decreasing the number of pedestrians that remain in the crosswalk at the release of conflicting traffic. Pedestrians appear to be adjusting their walking speed to finish crossing during the FDW interval. The countdown signals have not had the potentially negative effect of increasing the proportion of pedestrians entering the crosswalk during the FDW interval, for those pedestrians that also arrived at the crosswalk during the FDW. Overall, the countdown signals did not have a negative effect on pedestrian behavior such as running, hesitating and going back to point of start. In addition, the overall proportion of conflicts with vehicles decreased after the installation of the countdown signals. x

11 CHAPTER 1 INTRODUCTION The National Center for Statistics and Analysis reports that in the United States on average a pedestrian is killed every 111 minutes, and that every 8 minutes a pedestrian is injured in a crash, with 21 percent of fatalities occurring at intersections. In 2003, 4,288 pedestrians were killed and 68,000 pedestrians were injured in a crash. Improper crossing of an intersection is the main factor in pedestrian fatalities, followed by failure to yield the right of way and darting or running into the roadway. Florida ranked second in 2003 in the number of pedestrians killed, with a rate of 2.94 per 100,000 persons. The national rate for the same period was 1.63 per 100,000 persons (1). A common underlying problem that affects pedestrian safety at intersection crossings is the lack of understanding of the meaning of the pedestrian signal indications, particularly the Flashing UPRAISED HAND or FLASHING DON T WALK (FDW). The problems associated with the conventional pedestrian signal heads (Figure 1-1) are well documented in the literature: pedestrians are not sure about the meaning of the FDW indication; some pedestrians revert to their point of origin when the FDW indication is displayed; pedestrians illegally enter the crosswalk during the FDW, and with insufficient time to complete the crossing, are still in the crosswalk when the steady DW indication is displayed. The latter increases the potential for conflicts between pedestrians and vehicles, given that when pedestrians receive the DW indication, conflicting vehicle movements receive the green light to proceed. 1

12 2 Figure 1-1 Conventional Pedestrian Signal Indications. From Left: Steady DW and WALK. The clearance interval is indicated by the Flashing UPRAISED HAND or Flashing DON T WALK. Source: MUTCD 2003 (2) Chapter 4E. To address this problem and potentially increase the safety at signalized intersection crossings, the pedestrian countdown signal (Figure 1-2) was created by incorporating a countdown timer that is displayed simultaneously with the FDW indication. This timer displays the number of seconds remaining before the onset of the steady DW indication. It has been purported that this signal design leads to a higher level of pedestrian safety by enabling pedestrians to make better crossing decisions with the added information. The intended result is that a smaller percentage of pedestrians will remain in the crosswalk when the DW indication is displayed. Figure 1-2 Pedestrian Countdown Signal Indications. From Left: WALK, FDW and DW. The clearance interval displays the countdown timer concurrent with Flashing UPRAISED HAND or Flashing DON T WALK. Source: Photos of current installation at one of the study sites. The literature suggests that the compliance with the pedestrian signal indications may increase with the installation of pedestrian countdown signals. At the same time, some researchers have suggested that such signals may encourage more pedestrians to

13 3 enter during the FDW indication instead of waiting for the next WALK indication, by misjudging the time needed to safely complete the crossing. Due to considerable statewide interest in pedestrian safety from the traffic engineering community in Florida, the Florida Department of Transportation and the City of Gainesville sponsored this study of pedestrian countdown signals with the University of Florida Transportation Research Center. This study consists of observations of pedestrian behavior before and after the installation of pedestrian countdown signals at five intersections in Gainesville, Florida. The objective of this study was to evaluate the effect of pedestrian countdown signals on several performance measures, such as pedestrians compliance with the pedestrian signal indications and the percentage of pedestrians in the crosswalk during the DW interval, relative to the non-countdown pedestrian signals. The following tasks were conducted in support of this objective: A literature review was performed as the basis for comparison of similar studies and findings; Study sites were selected based on levels of pedestrian and vehicular traffic volumes, location, roadway geometry, and pedestrian population characteristics; Data collection was conducted using video cameras and signal status recording devices; The performance measures and methodology for data reduction were established; The data were reduced from the video tapes and coded into a spreadsheet format so that results could be summarized and analyzed; and, The data were analyzed using a population proportions (before and after installation of countdown signals) comparison test statistic for each selected performance measure. Chapter 2 consists of a literature review of regulations applicable to pedestrian signals as well as a comprehensive review of studies that investigate the applications and influences of both conventional pedestrian signals and countdown signals on pedestrian behavior and understanding of the signal indications. Chapter 3 presents the research

14 4 approach for this study, including detailed site descriptions and the methodology for data collection and data reduction. Chapter 4 discusses the selected performance measures for evaluation of the effectiveness of pedestrian countdown signals, the statistical analysis performed and the results obtained. Chapter 5 documents the findings and conclusions, and recommendations for further study.

15 CHAPTER 2 LITERATURE REVIEW A common underlying problem that affects pedestrian safety at crossings is the lack of understanding of the meaning of the pedestrian signal indications, particularly the Flashing UPRAISED HAND or FLASHING DON T WALK (FDW). Pedestrian countdown signals were created to address the problem. This chapter presents a review of the studies performed to date on the effectiveness of the implementation of pedestrian countdown signals as well as applicable regulations. Meaning of Pedestrian Signal Indications The 2001 edition of the Manual for Uniform Traffic Control Devices (MUTCD) (2) defines the meaning of the pedestrian signal indications as follows: - Steady WALKING PERSON (symbolizing WALK)... means that a pedestrian facing the signal indication may start to cross the roadway in the direction of the signal indication; - Flashing UPRAISED HAND (symbolizing DON T WALK)... means that a pedestrian shall not start to cross the roadway in the direction of the signal indication, but that any pedestrian who has already started to cross on a steady WALKING PERSON (symbolizing WALK) signal indication shall proceed out of the traveled way; - Steady UPRAISED HAND (symbolizing DON T WALK)... means that a pedestrian shall not enter the roadway in the direction of the signal indication. In the 2003 edition of the MUTCD (3) pedestrian countdown signals were included in the manual to facilitate the comprehension of the signal indications. A countdown timer was added to the FLASHING DON T WALK (FDW) interval to help inform pedestrians of the number of seconds remaining in the pedestrian change interval. In 5

16 6 April 2003 the Florida Department of Transportation (FDOT) added the application of pedestrian countdown signals to the Traffic Engineering Manual (4) to enhance the safety of the pedestrian crossing a street during the pedestrian phasing of the signal cycle. Florida law (5) states that pedestrians are subject to traffic signal control at intersections. It adopts the MUTCD as a standard regulation for traffic control devices, and states that pedestrians must conform to its requirements. Understanding of Pedestrian Signal Indications According to the FDOT State Safety Office (6), the majority of pedestrians who contact the office are found to be confused about the meaning of FDW indications. Pedestrians interpret this indication to mean that their allotted time to complete the crossing has ended and that vehicles will be given the green signal to resume movement. Robertson and Carter (7) found that only about half of the pedestrians understood the meaning of the FDW indication. Zegeer et al. (8) identifies the major pedestrian safety problem as the misunderstanding and confusion associated with pedestrian signal indications. Tidwell and Doyle (9) conducted a survey of over 4,800 people in 48 states to determine their level of understanding of pedestrian law and traffic control devices. The results indicated that nearly 50% of responders did not understand the meaning of the FDW indication, and that pedestrians tend to underestimate the ratio of pedestrian fatalities in relation to all traffic fatalities. Zegeer et al. (10) found that in some circumstances conventional pedestrian signal indications may create a false sense of security leading pedestrians to feel protected from vehicles and decrease their caution

17 7 while crossing. One third of the accidents observed by Zegeer et al. were caused by illegal pedestrian crossings. Khasnabis et al. (11) found that pedestrians tend to ignore pedestrian signal indications under low traffic volume conditions, and that the compliance with flashing signals tend to be lower than for steady signals. Yauch and Davis (12) relate the problems of lack of compliance with pedestrian indications to the continuing changes in design and operation associated with the pedestrian signals, which generate confusion and distrust. Conventional vs. Countdown Pedestrian Signal Indications A survey of 1,918 pedestrians conducted for the Federal Highway Administration (13) found that 80 percent of the pedestrians surveyed both before and after the installation of the countdown devices did not understand the meaning of the FDW indication. The countdown signals installed did not improve the pedestrians understanding of this indication. However, pedestrians felt safer at the crossings where the countdown devices were installed. Ullman et al. (14) compared five different treatments installed at seven intersections. Pedestrians were surveyed at each site. One of the sites included a pedestrian countdown signal. At this site 68 percent of pedestrians surveyed indicated that they felt the countdown signal was an effective crossing treatment and that it enhanced pedestrian safety at crossing, while 30 percent of respondents indicated that they did not understand the meaning of the countdown interval and felt unsure about how to proceed in the crossing due to turning vehicles. Botha et al. (15) conducted a before-and-after study of pedestrian countdown signals in the City of San Jose, CA. As part of the study he surveyed the pedestrians understanding of the FDW indication and found that misunderstanding of the FDW indication increases with the installation of countdown devices. Eighty percent of the

18 8 pedestrians surveyed indicated that they believed they could enter the crosswalk during the FDW indication if they were able to complete the crossing before the countdown reached zero. This assumption may have undesirable implications as the study also found that pedestrians were not able to correctly estimate the time needed to cross an intersection. Eccles et al. (16) conducted a similar before-and-after study of five intersections in Montgomery County, MD, and found that 62.6 percent of the pedestrians surveyed correctly interpreted the meaning of the countdown signals to be the remaining time for completing the crossing. Pulugurtha and Nambian (17) also found that understanding of the FDW interval increased after the installation of pedestrian countdown signals. Ninety percent of pedestrians surveyed correctly responded that the countdown timer indicates the time remaining to complete the crossing, and 85 percent associating the countdown signals with enhanced pedestrian safety at crossings. Influence on Pedestrian Behavior Eccles et al. (16) concluded that the pedestrian countdown signals had a positive effect on pedestrian behavior. While 2 of the 20 crosswalks evaluated resulted in a decrease in the number of pedestrians entering during the WALK interval, 6 experienced a significant increase. In addition, none of the intersections had a significant increase in the number of phases with pedestrians remaining in the crosswalk at the release of conflicting traffic, indicating that pedestrians adjusted their walking speed to clear the intersection faster even if entering the crosswalk later in the cycle. This study also found a significant decrease in the number of conflicts between pedestrian and motor vehicles after the installation of the countdown signals indicating that the use of such devices may increase pedestrian safety.

19 9 Huang and Zegeer (18) used a treatment and control study design to evaluate the effectiveness of pedestrian countdown signals in enhancing pedestrian compliance and understanding of the signals. The results indicate that the devices had a positive effect in reducing the number of pedestrians that started running once the FDW indication started. This study corroborates the findings of Eccles et al. in that the installation of countdown signals did not increase the number of pedestrians remaining in the crosswalk at the release of conflicting traffic, also citing the increase in walking speed to complete the crossing. The study, however, states that the compliance with the WALK indication decreased at the treatment sites. The authors did not recommend the installation of pedestrian countdown signals at standard intersections in Florida. It is noted that as a treatment and control site study, the results obtained may have been influenced by outside characteristics other than the treatment itself, such as intersection design and pedestrian population characteristics. Furthermore, the camera installation for data collection, at ground level, may have influenced pedestrian behavior. A Dutch review of pedestrian safety studies showed that on average only 35 percent of pedestrians cross during the WALK interval and that the type of destination had no impact on the probability of pedestrians crossing during the DW indication (19). It also mentioned a type of installation where the steady DW indication was replaced by a flashing indication which allows pedestrians to cross but at the same time alerts them that there could be traffic; crossing is at your own risk. The authors point out that the WALK indication does not necessarily mean a vehicle-free period as turning movements are frequently allowed during the pedestrian interval. The flashing indication is thought to improve safety by establishing a clear relationship between vehicles and pedestrians.

20 10 A study on the effects of such indications showed that the number of crossings outside of the WALK interval doubled, the waiting time for pedestrians was reduced and there was no increase in the number of pedestrian-vehicle conflicts. The argument on the potential for conflicts during the WALK interval is also found in Zegeer et al. (10) stating that the majority of crashes between pedestrians and vehicles at intersections involve a turning vehicle. Botha et al. (15) found, for the countdown signals, that the number of pedestrians entering the crosswalk during the FDW interval increased while the number exiting during the DW interval decreased, indicating that pedestrians may be adjusting their walking speeds to clear the intersection before traffic receives the green indication. At the same time, the compliance with the FDW indication has decreased, as a higher percentage of pedestrians that arrive during this interval enter the crosswalk instead of waiting for the next WALK interval. The authors conclude that although the countdown signals may have the undesirable effect of encouraging more people to enter the crosswalk during the FDW it also causes them to complete the crossing before the release of on-coming traffic, therefore reducing the chances of a conflict. No significant changes were observed in erratic pedestrian behavior such as running, stopping or hesitating. Pedestrian behavior at the crossing and willingness to comply with the pedestrian signal indication is also influenced by pedestrian delay caused by the signal as a function of timing, by the volumes of pedestrian and vehicular traffic, and by roadway characteristics such as width. According to Zegeer et al. (10) pedestrians that are willing to trust their own judgment of gaps in traffic incur less delay than those who comply with the signal.

21 11 FDOT s Traffic Engineering Manual (4) states that providing additional pedestrian clearance time information [using pedestrian countdown signals] will help the pedestrian decide whether to start the crossing or wait for the next WALK indication. Botha et al. (15) finds that this statement is correct, as the countdown signals provide additional information, and help in the pedestrian decision-making process. The noted reduction in the proportion of pedestrians remaining in the crosswalk during the DW interval was greater than the increase in the proportion of pedestrians entering during the FDW interval and can be considered as a positive effect. None of the studies reviewed have found an effect either positive or negative in the crash rate between pedestrians and motor vehicles after the installation of pedestrian countdown signals.

22 CHAPTER 3 RESEARCH APPROACH This chapter presents a detailed overview of the research methods and techniques used to evaluate the effectiveness of the pedestrian countdown signals using a beforeand-after study approach. Site Selection Five intersections in the City of Gainesville, Florida, were selected for the installation of pedestrian countdown signals. The selected intersections are located in the downtown area and in the vicinity of the University of Florida Campus. The user population at such intersections is representative of the City s population, with the exception of school-age children. Also, given the young profile of the City, the observations in this study do not represent the elderly segment of the population. The selected intersections are: 1. E University Avenue and E 1 st Street; 2. W University Avenue and W 2 nd Street; 3. W University Avenue and W 17 th Street; 4. W University Avenue and North-South Drive; and, 5. Archer Road and SW 16 th Street. All of the above intersections have high pedestrian traffic volumes during certain times of the day. Pedestrian counts will be presented later in the study. The vehicular traffic volumes are also high (annual average daily traffic above 20,000 veh/day) on the main approaches as illustrated by Figure 3-1 below. 12

23 NOTE: Drawing not to scale. Figure 3-1 Schematic of Count Station Locations SOURCE: City of Gainesville Public Works Department and Florida Department of Transportation Annual Average Daily Traffic Reports. Count data based on 2003 report. All of the study intersections are equipped with pedestrian push buttons with the exception of the minor approaches on E 1 st Street, W 2 nd Street and W 17 th Street. The pedestrian crossing cycles are push button-actuated, and are activated once per cycle if actuated. In addition, all crossings are signalized following the MUTCD guidance with R10-4 signs. At the intersection of W 17 th Street, the R10-3b signs are also installed. This is an educational sign that is used to help pedestrians understand the crossing signal indications. See Figure 3-2 for a sample installation at the intersection of W 17 th Street. The following section provides a detailed description of each site. Site Description E University Avenue and E 1 st Street This intersection is located in the downtown area and is surrounded by civic buildings such as City Hall, the County Administration Building, and the Civil Courthouse among others (see Figure 3-3). In addition, the downtown plaza and the main

24 14 bus transfer station are immediately adjacent to this intersection. The pedestrian population at this site is very diverse, composed of a mix of professionals (e.g., lawyers, engineering and financial consultants, etc.), retail and restaurant employees and patrons, students, and other miscellaneous users that visit the downtown area. R10-3b R10-4 Push Button Figure 3-2 Signage for Pedestrian Crossings W University Avenue and W 2 nd Street This intersection is also located in the downtown area, a few blocks to the west of the previous site (see Figure 3-3). This site has lower pedestrian traffic volume during daytime but a higher volume during nighttime, particularly on Thursday, Friday and Saturday nights, due to the presence of several nightclubs and bars in the surrounding area. W University Avenue and W 17 th Street The intersection of W University Avenue and NW 17 th Street is immediately adjacent to the University of Florida campus (see Figure 3-4). Other surrounding uses

25 15 include retail and restaurant areas. The volume of pedestrian traffic at this location is high throughout the day and consists primarily of students, staff and faculty. The user population at this site is composed primarily of university students, faculty and staff. Figure 3-3 Intersections in the Downtown Area: E University Ave and E 1 st St (high daytime activity level) and W University Avenue and W 2 nd St (high nighttime activity level) W University Avenue and North-South Drive This site is also adjacent to the University of Florida campus (see Figure 3-4). The pedestrian traffic level at this intersection is not as high as the intersection of NW 17 th Street mainly due to the surrounding uses, which are primarily residential with the exception of the University of Florida Foundation office and Alumni Hall, whose offices are on the north side of University Avenue. This intersection is heavily traveled during special events at the university (e.g., sporting events, graduation, concerts, etc.) due to the close proximity of stadium, gymnasium and parking areas. Archer Road and SW 16 th Street This intersection is immediately adjacent to, and provides direct access to Shands Hospital at the University of Florida. The pedestrian population at this site is composed mainly of hospital staff (e.g., doctors, nurses, janitorial, etc). Patients may also use this

26 16 intersection but it is not the primary access point for patients and visitors. Figure 3-5 below illustrates the intersection and surrounding area. Figure 3-4 Intersections adjacent to UF campus: W University Ave & 17 th St & N/S Dr Figure 3-5 Intersection of Archer Rd & SW 16 th St Data Collection Once the study sites were determined, the next step was to proceed with data collection. The data collection system used in this study is capable of simultaneously

27 17 capturing pedestrian and vehicular movements with a video camera and the corresponding traffic signal indications (20). The video cameras were mounted on signal poles at a specific location at each intersection as shown in Figure 3-6. Figure 3-6 Location of Camera Installation and Field of View at Each Study Intersection

28 18 Figure 3-6 Continued The videotapes collected from the field were then processed in the lab to obtain a composite video image of all vehicular and pedestrian movements along with the traffic

29 19 signal status for all recorded phases. A snapshot of this composite image, from W University Avenue and W 17 th Street, can be seen in Figure 3-7. The bars in the image screen indicate the signal phase interval for each approach and movement (either red or green). The ellipses correspond to the pedestrian signal indication: red for DW, yellow for FDW and green for WALK. Careful observation allows for the determination of pedestrian arrival time at the intersection, time of entry and exit at the crosswalk, any unusual pedestrian behavior due to the pedestrian signal indication, any conflicts between pedestrians and vehicles, the direction of movement, and in some instances, if the crossing was activated by the push of the pedestrian button. Figure 3-7 Composite Image Capture (W University Ave & NW 17 th St) The data collection system used in this study has several advantages: first, it is virtually undetectable by the subjects as all components but the video cameras are housed in the signal controller cabinet and the video camera (approximately 4 inches in length

30 20 and 1 inch in diameter) is mounted well above ground. Second, the recorded composite images allow for accurate and efficient data reduction. Data were recorded for several periods before and after the installation of the countdown signals. The before data were collected during the period from 9/30/2003 to 11/1/2003. The after data were collected during the period from 11/17/2003 to 3/24/2004 and 3/24/2004 to 4/15/2004. Data used for analysis in this study are shown in Table 3-2. The pedestrian countdown signals were installed between 10/28/2003 and 11/04/2003. In general, a minimum period of two weeks was allowed between the installation of the pedestrian countdown signals at each site and the beginning of the data collection for the after period of this study. A public education campaign did not accompany the activation of the countdown signals. Data Reduction For each intersection, data for one crosswalk crossing the major street and one crosswalk crossing the minor street were recorded. Periods of peak pedestrian traffic volume were selected for data reduction, typically: 7:00 a.m. to 9:00 a.m., 11:00 a.m. to 1:00 p.m., and 4:00 p.m. to 6:00 p.m. Also, for one intersection a late data collection period was used (10:00 p.m. to 2:00 a.m.) due to its vicinity to nightclubs and heavy pedestrian volume at nighttime (refer to Table 3-1). Data reduction from the videotapes was performed using an Excel spreadsheet (see Appendix A for sample data collection sheet). The following events were manually collected from each videotape and recorded in the spreadsheets: Time of pedestrian arrivals at the curb; Pedestrian signal indication at the time of pedestrian arrival; Whether the pedestrian used the pedestrian signal actuation button (if present);

31 21 Phase during which the pedestrian entered the crosswalk (W, FDW or DW); Cycle at which the pedestrian entered the crosswalk (same as arrival or following); Phase during which the pedestrian exited the crosswalk (W, FDW or DW); Erratic pedestrian behavior during crossing (hesitating, running or going back to starting point) due to the signal indication; and, Pedestrian-vehicle conflicts (stop, run, evade or crash).

32 22 Table 3-1 Data Collection Periods Used for Observation and Analysis SITE Data Collection: BEFORE Data Collection: AFTER E University Ave & E 1st St W University Ave & W 2nd St W University Ave & W 17th St W University Ave & W N/S Dr Archer Rd & SW 16th St 10/03/03 Friday 10/07/03 Tuesday 10/13/03 Monday 10/02/03 Thursday 10/07/03 Tuesday 10/17/03 Friday 10/18/03 Saturday 10/9/03 Thursday 10/10/03 Friday 10/13/03 Monday 10/14/04 Tuesday 10/01/03 Wednesday 10/07/03 Tuesday 10/08/03 Wednesday 10/09/03 Thursday 10/07/03 Tuesday 10/09/03 Thursday 10/10/03 Friday 10/13/03 Monday 7:00 AM - 9:00 AM 7:30 AM - 8:45 AM 4/02/04 11:00 AM - 2:00 PM Friday 11:45 AM - 1:00 PM 3:00 PM - 6:00 PM 4:30 PM - 6:00 PM 7:00 AM - 9:00 AM 7:30 AM - 8:45 AM 4/13/04 11:00 AM - 2:00 PM Tuesday 11:45 AM - 1:00 PM 3:00 PM - 6:00 PM 4:30 PM - 6:00 PM 7:00 AM - 9:00 AM 7:30 AM - 8:45 AM 4/14/04 11:00 AM - 2:00 PM Wednesday 11:45 AM - 1:00 PM 3:00 PM - 6:00 PM 4:30 PM - 6:00 PM 4/19/04 Monday TOTAL HOURS: 24 TOTAL HOURS: 16 7:00 AM - 9:00 AM 11:00 AM - 2:00 PM 3:00 PM - 6:00 PM 7:00 AM - 9:00 AM 12/05/03 Friday 12/06/03 Saturday 7:30 AM - 9:00 AM 11:30 AM - 12:50 PM 4:40 PM - 6:00 PM 10:00 PM - 2:00 AM 10:00 PM - 2:00 AM 11:00 AM - 2:00 PM 4/03/04 11:30 AM - 12:50 PM 3:00 PM - 6:00 PM Saturday 4:40 AM - 6:00 PM 10:00 PM - 2:00 AM 10:00 PM - 2:00 AM 4/05/04 Monday 4/13/04 Tuesday TOTAL HOURS: 24 TOTAL HOURS: 24 2:00 PM - 6:00 PM 3:00 PM - 6:00 PM 7:00 AM - 9:00 AM 11:00 AM - 2:00 PM 3:00 PM - 6:00 PM 11/17/03 Monday 11/18/04 Tuesday 11/19/04 Wednesday 12:30 PM - 1:50 PM 5:40 PM - 7:00 PM 7:40 AM - 9:00 AM 11:30 AM - 12:50 PM 4:40 AM - 6:00 PM 7:00 AM - 9:00 AM 11:00 AM - 2:00 PM 3:00 PM - 6:00 PM 11:00 AM - 2:00 PM 3:00 PM - 6:00 PM 7:00 AM - 9:00 AM 7:00 AM - 9:00 AM 3/26/04 11:00 AM - 2:00 PM 11:00 AM - 2:00 PM Friday 3:00 PM - 6:00 PM TOTAL HOURS: 21 TOTAL HOURS: 16 10:30 AM - 6:30 PM 7:30 AM - 8:45 AM 12/09/03 - Tuesday 4:30 PM - 6:00 PM 12/10/03 Wednesday 7:30 AM - 8:45 AM 11:45 AM - 1:00 PM 11:45 AM - 1:00 PM 7:30 AM - 8:45 AM 3/24/04 4:30 PM - 6:00 PM Wednesday 11:45 AM - 1:00 PM 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM 4:30 PM - 6:00 PM 7:30 AM - 8:45 AM 3/25/04 7:30 AM - 8:45 AM Thursday 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM TOTAL HOURS: 16 TOTAL HOURS: 12 6:45 AM - 8:00 AM 6:45 AM - 8:00 AM 3/25/04 11:45 AM - 1:00 PM Thursday 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM 4:30 PM - 6:00 PM 6:45 AM - 8:00 AM 6:45 AM - 8:00 AM 3/26/04 11:45 AM - 1:00 PM Friday 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM 4:30 PM - 6:00 PM 6:45 AM - 8:00 AM 6:45 AM - 8:00 AM 4/06/04 11:45 AM - 1:00 PM Tuesday 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM 4:30 PM - 6:00 PM 6:45 AM - 8:00 AM 11:45 AM - 1:00 PM 4:30 PM - 6:00 PM TOTAL HOURS: 16 TOTAL HOURS: 12

33 CHAPTER 4 RESULTS AND ANALYSIS This chapter presents the findings of the research. A total of 7,639 pedestrians were observed before and 6,339 pedestrians were observed after the installation of the pedestrian countdown signals. The observations recorded are shown in Tables 4-2 to 4-6. The results for each performance measure are analyzed using a test for difference in population proportions to evaluate if a significant difference between the before and after measurements can be attributed to the installation of the pedestrian countdown signals. Performance Measures From the reduced data described in Chapter 3, a number of performance measures were calculated for this study (for both before and after periods), including: The percentage of pedestrians entering the crosswalk during each signal indication (W, FDW and DW); The percentage of pedestrians exiting the crosswalk during each signal indication (W, FDW and DW); Compliance with the FDW indication; Percentage of pedestrians hesitating, running or going back to the point of start; and, Percentage of pedestrian-vehicle conflicts. Based on the assigned performance measures, the following questions were posed to evaluate the effectiveness of the pedestrian countdown signals: Are pedestrians more or less likely to comply with the W indication? Are pedestrians more or less likely to comply with the FDW indication? Are pedestrians more or less likely to be in the crosswalk during the DW indication when conflicting vehicle movement receives the green light? Did the occurrence of erratic behavior increase or decrease with the installation of countdown signals? 23

34 24 Did the occurrence of pedestrian-vehicle conflicts increase or decrease with the installation of countdown signals? Statistical Analysis To test the significance of the results observed, a test for the difference in population proportions was used (21). This test was performed to evaluate if the changes in the performance measures observed between the before and after measurements are statistically significant, indicating that the pedestrian countdown signals have influenced pedestrian behavior. The hypothesis testing is based on the z statistic from a normal distribution. The calculations were performed using the following formula: z = pˆ ( 1 pˆ ) pˆ ( 1 p ) ˆ2 n p 1 p ˆ1 ˆ 2 + n 2 where z = calculated test statistic, z α / 2 = critical z value from table of normal distribution probabilities for a given confidence level, ˆp 1 = estimate of before population proportion (for specified performance measure), ˆp 2 = estimate of after population proportion (for specified performance measure), n 1 = before sample size, and n 2 = after sample size. The null hypothesis tested in all cases is that there is no difference between the before and after proportions, with the alternate hypothesis that there is a difference. The rejection region (R.R.) for the statistical test is defined for a two-tailed z test as follows:

35 25 H H o a : pˆ 1 : pˆ 1 pˆ R. R.: z 2 pˆ 2 z α / 2 = 0 0 A two-tailed z test was performed at a confidence level of 95%. The critical z value (z α/2 ) obtained from the table of normal distribution probabilities for the given confidence level is The calculated z values for each performance measure are shown in Table 4-1 below. These values are compared to the critical z value. The null hypothesis that the proportion values observed during the before and after periods are equal is rejected if the absolute calculated z value is higher than the critical z value. Positive z-values indicate that the proportion observed in the after period has decreased for the particular performance measure. The reverse is true for negative values, which indicate that the proportion observed during the after period has increased. A discussion of findings is presented in the following section. Discussion of Results by Performance Measure Pedestrians Entering the Crosswalk The number of pedestrians entering the crosswalk during each of the pedestrian phase intervals (WALK, FDW, DW) was recorded during the before and after phase of the study for peak periods of pedestrian activity as indicated in the previous chapter. The proportions for each performance measure are shown in Table 4-2. There was a significant increase in the proportion of pedestrians entering the crosswalk during the WALK indication and a significant decrease in the proportion of pedestrians entering during the DW indication at 3 of the 5 study intersections. As for the proportion of pedestrians entering the crosswalk during the FDW, there was a significant increase

36 26 observed at the intersection of Archer Road and SW 16 th Street; and a significant decrease at the intersection of E University Avenue and E 1 st Street. No significant changes were observed for pedestrians entering during FDW at the remaining 3 intersections. Overall, the installation of pedestrian countdown signals had a positive effect on the behavior of pedestrians entering the crosswalk. The findings are discussed below. Table 4-1 Calculated Test Statistic (z value) by Performance Measure Event Intersection Overall Exit-FDW Exit-DW Enter-W Enter-FDW Enter-DW Compliance FDW Wait for WALK Non-Compliance Enter at FDW Non-Compliance Enter at DW Hesitating Running Going Back Conflict - Run Conflict - Stop Conflict - Evade Note: 1. Intersections are: 1) E University Ave and E 1 st St; 2) W University Ave an d W 2 nd St; 3) W University Ave and W 17 th St; 4) W University Ave and N/S Dr; and, 5) Archer Rd and SW 16 th St. 2. Bold values are statistically significant at the 95% confidence level. Pedestrians entering at WALK. The proportion of pedestrians entering the crosswalk during the WALK indication significantly increased after the installation of the pedestrian countdown signals at 3 of the study intersections: W University Avenue and W 17 th Street, W University Avenue and North-South Drive, and Archer Road and SW

37 27 16 th Street. It significantly decreased at the intersection of E University Avenue and E 1 st Street, and no significant change occurred at the intersection of W University Avenue and W 2 nd Street. Table 4-2 Pedestrians Entering Crosswalk SITE TOTAL ENTERING AT "W" ENTERING AT "FDW" ENTERING AT "DW" Before After Before After Diff. Before After Diff. Before After Diff. EUA & E 1st ST % 34.73% % 13.99% 6.39% -7.60% 34.28% 58.48% 24.20% WUA & W 2nd ST % 33.74% 1.31% 9.41% 11.34% 1.93% 58.16% 54.93% -3.23% WUA & W 17th ST % 40.74% 13.71% 8.14% 9.27% 1.13% 64.83% 49.98% % WUA & N/S DR % 85.71% 9.92% 8.56% 6.56% -2.00% 15.65% 7.72% -7.93% ARCHER & SW 16th AVE % 69.41% 15.75% 13.35% 17.37% 4.02% 32.98% 13.22% % Pedestrians entering at FDW. The proportion of pedestrians entering the crosswalk during the FDW interval increased at only one of the study intersections Archer Road and SW 16 th Street. At the intersection of W University Avenue and E 1 st Street there was a decrease, while at all other intersections no statistically significant change was noted. Pedestrians entering at DW. One of the most significant performance measures is the number of pedestrians that enter the crosswalk illegally during the DW indication, as this has the highest potential for a conflict between pedestrians and vehicles. The study results show that the proportion of pedestrians entering the crosswalk during the DW significantly decreased at 3 of the sites after the installation of the pedestrian countdown signals: W University Avenue and W 17 th Street, W University Avenue and North-South Drive, and Archer Road and SW 16 th Street. At the intersection of E University Avenue and E 1 st Street the proportion increased significantly, while at W University Avenue and W 2 nd Street no significant change occurred.

38 28 At the intersection of W University Avenue and W 2 nd Street, the peak period of pedestrian activity is during the nighttime, particularly between 1:30 a.m. and 2:00 a.m. when the bars and nightclubs close and customers must leave. At this time of night there is a general disregard for the pedestrian signal indications. First, there is a good chance that pedestrians (mostly college students) are under the influence of alcohol, and second because large groups of people attempt to cross at the same time and vehicles tend to yield the right-of-way. It should be noted that the results observed at the intersection of E University Avenue and E 1 st Street may have been influenced by signal timing changes that decreased the length of time allowed for the pedestrian crossings. The cycle length at this intersection is very high during the peak times of observation. During mid-day the traffic volume is lower, and pedestrians tend to enter and exit during the DW interval taking advantage of gaps in traffic for completing the crossing. In some instances pedestrians cross to the mid point of the roadway and wait at the centerline for a gap to finish the crossing. Furthermore, pedestrians do not always press the push button for actuation of the pedestrian cycle, generating longer periods of delay. Another factor that influences the differences in behavior at this site is the presence of the Regional Transit System (RTS) downtown bus transfer station located in close proximity to this intersection. A high volume of pedestrians arrive at the intersection after getting off a bus. Pedestrians also cross University Avenue to get to the transfer station. There was little compliance with the pedestrian signal indication from pedestrians arriving at the intersection coming from or going to the bus station.

39 29 Pedestrians Exiting the Crosswalk The number of pedestrians exiting the crosswalk was recorded. The proportion of pedestrians exiting the crosswalk during the FDW indication significantly increased at 4 of the 5 study sites after the installation of the pedestrian countdown signals, while the proportion exiting during the DW indication significantly decreased at 3 of the 5 study sites. The proportion exiting during the WALK interval is not considered as a useful indication as pedestrians finishing during this interval are most likely completing the crossing after starting during the DW indication. In addition, pedestrians are not expected to finish during this interval as it is timed for only a few seconds allowing the pedestrian to reach a certain distance but not to finish the crossing. The desired outcome is for the pedestrians to finish during the FDW indication, clearing the intersection prior to conflicting vehicles receiving the green light. The proportion of pedestrians exiting during each interval is shown in Table 4-3. Table 4-3 Pedestrians Exiting Crosswalk SITE TOTAL EXITING AT "W" EXITING AT "FDW" EXITING AT "DW" Before After Before After Diff. Before After Diff. Before After Diff. EUA & E 1st ST % 2.79% -3.77% 61.26% 35.93% % 32.18% 61.28% 29.10% WUA & W 2nd ST % 6.88% -7.21% 26.92% 36.71% 9.79% 59.00% 56.41% -2.59% WUA & W 17th ST % 4.40% -0.37% 43.64% 53.61% 9.97% 51.60% 41.98% -9.62% WUA & N/S DR % 1.16% 0.18% 77.75% 88.42% 10.67% 21.27% 10.42% % ARCHER & SW 16th AVE % 8.45% 0.44% 58.32% 72.85% 14.53% 33.66% 18.70% % Pedestrians exiting on FDW. There was a significant increase in the proportion of pedestrians exiting the crosswalk during the FDW indication after the installation of the pedestrian countdown signals. This behavior is observed at all study intersections with the exception of the intersection of E University Avenue and E 1 st Street. The increase in

40 30 exits during the FDW may be a result of pedestrians taking advantage of the information given by the countdown and adjusting their walking speed to complete the crossing prior to the release of conflicting vehicular traffic. Pedestrians exiting on DW. The proportion of pedestrians exiting the crosswalk during the DW indication significantly decreased after the installation of the countdown signals at 3 of the study intersections: W University Ave and W 17 th Street, W University Avenue and North-South Drive, and Archer Road and SW 16 th Street. There was no significant change at the intersection of W University Avenue and W 2 nd Street. At the intersection of E University Avenue and E 1 st Street there was an increase in the number of pedestrians exiting during the DW interval. The latter may also be associated with unique conditions present at this intersection as discussed in the previous section. Compliance with FDW Indication In the literature there is extensive documentation of pedestrians misunderstanding of and lack of compliance with, the FDW indication. In this study, compliance with the FDW indication is measured by the behavior of pedestrians arriving at the intersection during the FDW. A pedestrian complies with the FDW indication if upon arrival during the FDW, (s)he stops and waits for the next WALK interval during the next cycle. Noncompliance is measured by (s)he entering the crosswalk during the FDW or DW indication after arriving during the FDW. Of the total arrivals observed over the course of this study, 8.5% (655) of the pedestrians arrived during the FDW on the before phase, and 9.4% (598) of the pedestrians arrived during the FDW on the after phase. The results indicate that there was no statistically significant difference in the compliance of pedestrians with the FDW indication between the before and after installation of the pedestrian countdown signals, except at the intersection of W

41 31 University Avenue and W 2 nd Street. At this intersection there was an increase in the proportion of pedestrians that complied with the FDW indication by waiting for the next WALK indication and refraining from entering during the FDW. The proportions observed are shown in Table 4-4. Table 4-4 Compliance with the Flashing Don t Walk Indication SITE ARRIVALS AT "FDW" COMPLIANCE NON-COMPLIANCE WAITING FOR "W" ENTERING AT "FDW" ENTERING AT "DW" Before After Before After Diff. Before After Diff. Before After Diff. EUA & E 1st ST % 3.85% -0.96% 87.50% 88.46% 0.96% 7.69% 7.69% 0.00% WUA & W 2nd ST % 12.24% 12.24% 98.77% 86.73% % 1.23% 1.02% -0.21% WUA & W 17th ST % 1.90% -0.17% 91.70% 90.87% -0.83% 6.22% 7.22% 1.00% WUA & N/S DR % 6.25% 2.55% 92.59% 93.75% 1.16% 3.70% 0.00% -3.70% ARCHER & SW 16th AVE % 14.87% -3.94% 71.78% 76.92% 5.14% 9.41% 8.21% -1.20% Erratic Pedestrian Behavior Pedestrian behavior as a function of the pedestrian signal indication was recorded and analyzed during the before and after phases of this study. The erratic behaviors observed are defined as: Pedestrian hesitates: pedestrian is unsure if (s)he can start the crossing. This is observed by pedestrian initiating the movement by stepping off the curb and waiting in the crosswalk before deciding to proceed and complete the crossing; Pedestrian runs: pedestrian starts to run when the pedestrian signal indication changes or as timer during the FDW approaches zero; and, Pedestrian goes back to point of start: pedestrian initiates the crossing but turns around and returns to point of origin based on interpretation of the pedestrian signal. Pedestrians that crossed diagonally and jogging were not included in the analysis. In addition, pedestrians that crossed outside of the marked crosswalk, jaywalking, were considered to be outside of the field of influence of the pedestrian signals, and were also