Appendixes to TCRP RRD 84: Audible Signals for Pedestrian Safety in LRT Environments (2007)

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.

Appendix D Field Test Report (Prepared February 2007)

Field Test Report Table of Contents FIELD TEST......................................................................................................................1 BACKGROUND ................................................................................................................1 BASE CASE CONDITIONS ..............................................................................................1 TESTED ALTERNATIVES................................................................................................2 FIELD OBSERVATIONS...................................................................................................4 Training Data Collectors ................................................................................................4 General Scoring Procedure ...........................................................................................5 Measures.......................................................................................................................5 EXPERIMENTAL DESIGN................................................................................................6 FIELD OBSERVATION RESULTS ...................................................................................6 VISUAL IMPARED FIELD SURVEY...............................................................................13 PUBLIC SURVEYS.........................................................................................................15 CONCLUSIONS..............................................................................................................17

Field Test Report List of Figures Figure 1 Mid-Block Crosswalk...........................................................................................3 Figure 2 Visual train warning sign ....................................................................................4 Figure 3 The percent of violators during each treatment condition ..................................7 Figure 4 Percentage of violators looking for the train before crossing the track ..............8 Figure 5 Percent violators dashing across rails during each condition ............................9 Figure 6 The mean interval between violations and train arrival....................................10 Figure 7 Histograms showing the frequency of violations for each time period .............11 List of Tables Table 1 Proportion of Violators.........................................................................................7 Table 2 Violator Orientation Results ................................................................................8 Table 3 Proportion of Violators that Dash ........................................................................9 Table 4 Analysis of Variance for the Mean Interval Between Warning and Looking........9 Table 5 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train in Seconds ...............................................................................................................10 Table 6 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train Arrival at 10 Seconds or Less Between Baseline and Bell Alone ............................12 Table 7 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train Arrival at 7.5 Seconds or Less Between Baseline and Unique Sound ....................12 Table 8 Incidents of Evasive Conflicts During Each Scenario .......................................13 Table 9 Survey Responses to the Audible Signal ..........................................................16

Field Test Report FIELD TEST This report outlines the results of the Task 10 Field Test portion for the TCRP D-10 project entitled “Audible Signals for Pedestrian Safety in LRT Environments.” The field test included behavioral observations, a visually impaired field survey, and a public survey of two alternative audible warnings. Each audible warning was tested with and without a visual device and compared to the base case (existing) conditions at the crossing. The field test was located in downtown Salt Lake City, Utah at 50 South Main Street where both of Utah Transit Authority (UTA) light rail lines cross the crossing. The purpose of the field test was to determine if a new audible warning or a bell warning at pedestrian/rail crossings would improve pedestrian safety. Since occurrences of pedestrian/light rail vehicle accidents are rare the potential safety improvement was measured by observing changes in risky behavior. First, the background that led to the field test will be provided. The base conditions of the study area will then be discussed followed by the alternatives to be tested. The field tests conducted will then be discussed along with the results. Finally, conclusions will be provided with a summary of the overall results. BACKGROUND The scope of this project was to consider the development of an alternative audible warning. To this end, the auditory engineer developed and tested a unique sound in a research laboratory. Based on the laboratory results two warnings were selected for field testing: the conventional bell and the unique “blended staircase” signals. The unique “blended staircase” consists of two components: the familiar sound of an approaching train and a conventional crossing bell processed through a pitch-shifting algorithm. This unique audible warning was designed such that a pedestrian approaching a grade crossing successively hears both a bell-like sound of rising pitch and an approaching train of increasing loudness. This auditory icon provides more information for the same degree of annoyance. The unique sound and bell signal were evaluated and compared in the field trial in Salt Lake City. Prior research (Fidell, 1978; Fidell et al., 1979; Sneddon et al., 2004) has demonstrated that sounds of equal audibility, regardless of aversiveness, are equally effective in attracting attention for warning purposes. A description and the results of the field tests are provided in the remainder of this report. BASE CASE CONDITIONS This mid-block crossing is located immediately north of the City Center TRAX Station between two downtown malls. The City Center TRAX Station is one of UTA’s busiest stations. This location was selected because of the relatively high number of pedestrian/LRV conflicts. The area has pedestrian activity throughout the day. The base conditions included audible pedestrian signals in conjunction with pedestrian heads with countdown timers on each side of the crossing. The pedestrian signal stays

Field Test Report in the WALK position and turns to DON’T WALK only when an automobile or LRV approaches. Stenciled markings on the pavement that state, “Look Both Ways,” are located at each end of the crosswalk, as is common practice in the downtown Salk Lake City area The base conditions do not include an automated audible LRV crossing warning. There are several non-automated warnings. First, northbound trains stop at the City Center TRAX Station immediately before the pedestrian crossing. Before leaving the station and entering the pedestrian crossing the train operator will sound a gong twice. Second, southbound trains pass through the pedestrian crossing before stopping at the City Center TRAX Station. Trains will gong if a pedestrian is in the train’s path or near the tracks. Also, according to UTA’s TRAX Rule Book, in the case where a train is already at a station and a second train approaches the arriving train must reduce its speed to 10 mph and sound the gong continuously until the front of the train has safely cleared the rear of the other train (UTA rule 4.24). The operators first use the gong, and if the situation necessitates it, then they will use the louder horn. The horn may be sounded by the operator as a warning whenever an unsafe or emergency condition exists. TESTED ALTERNATIVES Three components of the field test were performed consisting of: • Behavioral observations, which were video taped, and then scored and analyzed; • A survey of visually impaired pedestrians; and • A public survey. This section provides the general setup under which the three tests were performed. The mid-block crosswalk is shown in plan view along with photos of the existing (base case) condition in Figure 1. The Audible Pedestrian Signal DS-200 Series manufactured by Novax Industrial Corporation were installed by Salt Lake City staff, and Siemens ITS programmed the device into the existing signal system. The speakers for the audible warning were located in two new pedestrian heads (one on each side of the crossing) that were mounted next to the existing pedestrian heads with countdown timers. The typical pedestrian face plates were replaced with a yellow activated LRT sign (W10-7). This visual sign was programmed to turn on and remain on constantly for the duration of the audible warning in the scenarios when the visual sign was used. The new activated LRT sign is shown in Figure 2 next to the existing pedestrian head with a countdown timer. In addition to the base case four treatment scenarios were explored: • Conventional bell without the visual sign, • Conventional bell with the visual sign, • Unique sound without the visual sign, and • Unique sound with the visual sign.

Field Test Report Figure 1 D-10 Field Test Plan Main Street Crosswalk

Field Test Report Figure 2 Visual train warning sign The conventional bell stays at the same pitch and is currently used at train crossings with automatic gates across the country including in the Salt Lake City area. The unique sound is the “blended staircase” discussed in the background section of this report. The visual impaired field survey did not consider the “with the visual sign” scenarios. However, a base case scenario, “Normal”, was added where only on-vehicle audible warnings were given as discussed in the base case conditions section of this report. FIELD OBSERVATIONS The observation test was completed by video taping pedestrian behavior at the LRT crossing. This provided a direct measure of the effects of each device on pedestrian behavior. At the proposed test site there are often groups of people crossing together and it would be difficult to collect some data such as the latency between the start of the audible device and specific pedestrian behaviors in real time. A fixed camera proved better for scoring latency because of the time code feature in the camera. Also from a video tape it was easier to score the behavior of individuals that are part of a large group, and provided an accurate record of all the data that was collected. Data was collected by video taping the crossing for three, eight-hour periods for each of the five test scenarios. Korve supervised all the data collection at the Salt Lake City test site. The data from the cameras was sent to Dr. Van Houten, who used research assistants to collect the pertinent data from the video tapes. The video camera was place on the east side of Main Street as shown in Figure 1. Training Data Collectors Data collectors were trained how to record each measure of effectiveness (MOE) by senior staff. Training continued until a measure of inter-observer agreement between observers was above 85%. Measures of inter-observer agreement were collected for a minimum of 20% of observations to assure that observers continued to use the same standards when observing target behaviors.

Field Test Report General Scoring Procedure All data were scored from videotape taken with a digital camera. The camera was located at grade level on the east side of the crosswalk facing the pedestrian signal on the west side of the crosswalk. The camera was set up so the pedestrian signal and train symbol signal were both clearly visible as well as the entire width of the crossing. Observers used fast forward to advance between trains and then rewound the tape until the start of the last WALK indication prior to the train’s arrival. This procedure was used in order to keep conditions comparable between audible warning and baseline conditions. In almost all cases when an audible warning was present, it began during the pedestrian clearance interval (flashing DON’T WALK) period. Pedestrians were considered in violation if they began to cross after the start of the flashing DON’T WALK signal. Violators could cross at any time prior to the train’s arrival blocking their crossing. Crossing outside the crosswalk was not scored because the location of the platforms prevented crossing outside the crosswalk within the view of the cameras. Because there were often more than one violator per crossing, it was necessary to repeat the scoring procedure for each violator. In general the portion of tape showing an individual train arriving or departing had to be replayed several times in order to accurately score data for all pedestrians. Measures - Did not violate Signal. The pedestrians were scored as being in compliance with the pedestrian control signal when he or she came to a complete stop and waited until the train has cleared the crossing before crossing the rails. - Pedestrian Violations. A pedestrian violation was scored whenever a pedestrian crossed the path of an approaching train while violating the pedestrian control signal indicated by the start of the flashing DON’T WALK signal (during treatment this was typically at the onset of the audible warning). Per this criterion, violators could cross at any time from the onset of the flashing DON’T WALK until a train blocked the crossing. It is important to note that this procedure does yield a higher level of non-compliance than one would obtain if one included all pedestrians including those that could not be influenced by the added audible warnings as it excludes pedestrians who crossed when trains were not approaching the crossing, regardless of whether they complied with the WALK/DON’T WALK indication. - Violator who stopped before crossing the rails. The pedestrian was scored for this behavior if he or she stopped before crossing the rail in violation of the pedestrian signal. The percentage of pedestrians that stopped during each condition was scored from videotape. A stop was defined as checking forward motion for a period of at least a second when no object or person was blocking the path of the pedestrian. - Violator slowed before crossing the rails. This behavior was scored if the pedestrian visibly slowed before crossing the rail while violating the pedestrian signal prior to the arrival of a train.

Field Test Report - Violator looked at approaching train. Looking behavior was defined as turning the head in the direction of the approaching train or glancing in the direction of the approaching train just prior to crossing the track (within 3 seconds of crossing the rails). - Look train latency. The time from the onset of the audible warning until the person looked in the direction of the train was scored from the videotapes. The average look latency was computed using a stopwatch. - Time between pedestrian violation and train arrival. The interval between a pedestrian crossing in violation of the crosswalk and arrival of the train at the location where the pedestrian crossed was computed using a stopwatch. - Train-pedestrian conflicts (train action). A train-pedestrian conflict was scored whenever the train motorman had to suddenly brake in order to avoid striking a pedestrian crossing in front of the train. For a conflict to be scored the person had to be on a collision course with the train before taking evasive action. - Train-pedestrian conflicts (pedestrian action). A train-pedestrian conflict was scored whenever a pedestrian had to take evasive action such as jumping back or running faster to avoid being struck by a train. For a conflict to be scored the person had to be on a collision course with the train before taking evasive action. EXPERIMENTAL DESIGN Baseline data, and treatment data were collected on different days. Data were often alternated between data collection between different conditions so that some baseline data were collected before the introduction of audible warnings and some baseline data were collected after the introduction of audible warnings. Because all data on the different audible warnings were collected at the same site, a characteristic of the particular site was not confounded with the type of device tested. This allowed a within comparison of the efficacy of each of the audible warnings with the baseline (no treatment) conditions. FIELD OBSERVATION RESULTS The percentage of violators during baseline and the bell and unique sound conditions are presented in Figure 3. The bell alone condition was associated with the highest percentage of pedestrians violating the signal (55.2%) and the other test conditions had a slightly lower percentage of pedestrians violating the signal (47.3% for the unique sound plus visual to 51.7% for the bell plus visual). Note these violation rates were with respect to the train approach periods only, and are significantly higher than the rate if all pedestrian activity was included. The majority of the pedestrians cross during the walk phase, and these pedestrians were not counted in this analysis because a train was not approaching so the audible warning device we were testing was not activated. Table 1 shows that the differences between these treatments were not statistically significant.

Field Test Report 0 10 20 30 40 50 60 70 80 90 100 Baseline Bell alone Bell + visual icon Unique sound + Visual icon Unique sound + Visual icon Condition Figure 3 The percent of violators during each treatment condition Table 1 Proportion of Violators Condition Description Baseline Bell Alone Bell Plus Visual Unique Sound Alone Unique Sound plus Visual Total n Non Violator 208 232 261 261 272 1,234 Violator 202 286 244 241 244 1,217 Total 410 518 505 502 516 2,451 % of Violators 49.2 55.2 51.7 48.0 47.3 A 2 x 5 Chi Square analysis revealed that there were no differences among the treatments, χ 2 (4, N = 2451) = 8.88, p = .075, there is not enough evidence to suggest a difference among these treatment conditions. The percentage of violators looking for the train just prior to crossing the tracks is shown in Figure 4. Looking was lowest in baseline (34.3% looked) and highest in the unique sound alone condition (42.6% looked), however the differences compared with a chi square test were not found to be significantly different from each other. The results of this analysis are presented in Table 2. The number of violators that dashed across the rails during each condition is shown in Figure 5 and Table 3. These data show that the percentage dashing ranged from 17.5% in the unique sound plus visual condition to 27.2% during the bell plus visual condition. The chi square analysis revealed a significant difference at the .05 level. A post hoc 2 x 2 Chi Square analysis revealed that there was a difference between the baseline and bell plus visual treatments, (1, N = 2451) = 8.054, p = .005 (Bonferroni corrected alpha = .045).

Field Test Report 0 10 20 30 40 50 60 70 80 90 100 Baseline Bell Bell + visual icon Unique. Sound Unique sound + visual icon Condition Figure 4 Percentage of violators looking for the train before crossing the track Table 2 Violator Orientation Results Condition Violator Visual Orientation Baseline Bell Alone Bell Plus Visual Unique Sound Alone Unique Sound plus Visual Total n Look 110 95 122 109 53 488 No Look 211 156 168 147 85 767 Total 321 251 290 256 137 1255 A 2 x 5 Chi Square analysis revealed that there were no differences among the treatments, χ 2 (4, N = 1255) = 5.74, p = .219, there is not enough evidence to suggest a difference among these treatment conditions. The percentage of pedestrians that stopped or broke stride before crossing the rails was consistently low during all conditions and the values were very similar ranging from 1.8% during the unique sound and 3.4% during the bell plus visual condition. A chi square test revealed that these values were not significantly different from each other. An examination of the interval between the warning and looking was also found to be insignificant.

Field Test Report 0 5 10 15 20 25 30 Baseline Bell Alone Bell + visual icon Unique. Sound Unique. sound + visual icon Condition Figure 5 Percent violators dashing across rails during each condition Table 3 Proportion of Violators that Dash Condition Violator Baseline Bell Alone Bell Plus Visual Unique Sound Alone Unique Sound plus Visual Total n Dash 57 54 79 61 24 275 No Dast 264 197 210 195 113 979 Total 321 251 289 256 137 1,254 A 2x5 Chi Square analysis revealed that there were differences among the treatments, χ 2 (4, N = 1254) = 10.319, p = .035, there is sufficient evidence to suggest a difference among these treatment conditions. Table 4 Analysis of Variance for the Mean Interval Between Warning and Looking Source Df MS F p Between Groups 1 11.497 .311 .578 Within Groups 172 36.997 Total 173 There is not enough evidence to suggest that there is a difference between the treatment conditions in terms of the mean interval between the type of warning and looking. The mean violation interval in seconds is presented in Figure 6. These data varied between a low of 15.8 seconds in the unique sound alone condition and a high of 16.8 seconds in the baseline condition. Theses data were analyzed using an ANOVA and no difference was found between the mean intervals for each condition.

Field Test Report 0 2 4 6 8 10 12 14 16 18 20 Baseline Bell Alone Bell + visual icon Unique sound Unique sound + visual icon Condition Figure 6 The mean interval between violations and train arrival Table 5 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train in Seconds Source Df MS F p Between Groups 4 49.597 1.121 .345 Within Groups 1249 44.252 Total 1253 There is not enough evidence to suggest that there is a difference among the treatments conditions in terms of the mean interval of violator crossing. Histograms showing the distribution of times between crossing and train arrival for thee conditions in the experiment are presented in Figure 7. Although there were no significant differences between the mean crossing times and train arrival between each of the five conditions, an examination of these distributions shows that relatively fewer pedestrians crossed within 10 seconds of the train arrival during the warning bell condition. The difference between the base condition and the two audible treatments, the bell and the unique sound, are both statistically significant (Table 6 and 7).

Field Test Report B a s e l in e In t e r v a l B e t w e e n P e d e s t r ia n V io la t io n & T r a in A r r iv a l ( s e c ) 3 0 . 0 2 7 . 5 2 5 . 0 2 2 . 5 2 0 . 0 1 7 . 5 1 5 . 0 1 2 . 5 1 0 . 0 7 . 5 5 . 0 2 . 5 B a s e l i n e I n t e r v a l B e t w e e n P e d e s t r i a n V i o l a t i o n & T r a i n A r r i v a l ( s e c ) Fr eq ue nc y 6 0 5 0 4 0 3 0 2 0 1 0 0 S t d . D e v = 6 . 9 6 M e a n = 1 6 . 8 N = 3 2 1 . 0 0 B e l l O n l y In t e r v a l b e t w e e n V i o l a t i o n & T r a i n A r r i v a l ( s e c ) 3 0 . 0 2 7 . 5 2 5 . 0 2 2 . 5 2 0 . 0 1 7 . 5 1 5 . 0 1 2 . 5 1 0 . 0 7 . 5 5 . 0 2 . 5 B e l l O n l y In t e r v a l b e t w e e n V i o l a t i o n & T r a i n A r r i v a l ( s e c ) Fr eq ue nc y 5 0 4 0 3 0 2 0 1 0 0 S t d . D e v = 6 . 3 7 M e a n = 1 6 . 0 N = 2 5 1 . 0 0 U n i q u e S o u n d A l o n e In t e r v a l B e t w e e n V i o l a t i o n & T r a n A r r i v a l ( s e c ) 3 0 . 0 2 7 . 5 2 5 . 0 2 2 . 5 2 0 . 0 1 7 . 5 1 5 . 0 1 2 . 5 1 0 . 0 7 . 5 5 . 0 2 . 5 U n i q u e S o u n d A l o n e In t e r v a l B e t w e e n V i o l a t i o n & T r a n A r r i v a l ( s e c ) Fr eq ue nc y 5 0 4 0 3 0 2 0 1 0 0 S t d . D e v = 6 . 3 8 M e a n = 1 5 . 8 N = 2 5 6 . 0 0 Figure 7 Histograms showing the frequency of violations for each time period

Field Test Report Table 6 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train Arrival at 10 Seconds or Less Between Baseline and Bell Alone Source Df MS F p Between Groups 1 21.75 4.217 .042 Within Groups 123 5.15 Total 124 There is enough evidence to suggest that there is a difference between the baseline and bell alone conditions in terms of the mean interval of violator crossing less then 10 seconds before the train arrives. Means: Baseline, 6.99 seconds and Sound Alone, 7.83. Table 7 Analysis of Variance for the Mean Interval of Violator Crossing Before the Train Arrival at 7.5 Seconds or Less Between Baseline and Unique Sound Source Df MS F p Between Groups 1 8.25 4.03 .048 Within Groups 69 2.04 Total 70 There is enough evidence to suggest that there is a difference between the baseline and unique sound conditions in terms of the mean interval of violator crossing less then 7.5 seconds before the train arrives. Means: Baseline, 5.67 seconds and Unique Sound, 4.96. The incidents of evasive conflicts were very rare in this study. The total of 5 evasive conflicts observed during this study are described here and summarized in Table 8. One occurred during baseline when a train had to brake suddenly to avoid hitting three pedestrians, one of who crossed 3 seconds prior to the train arrival and two of whom crossed 2 seconds before the train arrival. One conflict occurred during the bell alone condition when 2 pedestrians crossed 1 seconds before the train arrival. In this case 1 pedestrian looked and the other did not, both pedestrians lunged forward. Two conflicts occurred during the bell plus visual condition. Both conflicts involved the motorman having to suddenly brake and one pedestrian looked and the other did not. The times between crossing and train arrival for these two independent events were both 1 second. No conflicts occurred during the unique sound alone condition, and one evasive conflict occurred during the unique sound plus visual condition. This conflict involved the motorman suddenly braking and the pedestrian crossing 6 seconds before the train arrival after looking. Unfortunately it is not possible to draw any firm conclusions from this very small sample of conflicts. However, the individual conflicts were associated with short times between crossing and train arrival and all but one involved the motorman braking for the pedestrians. Because of the slow speeds involved it was relatively easy for the train operator to brake.

Field Test Report Table 8 Incidents of Evasive Conflicts During Each Scenario Baseline Bell No Visual Bell With Visual Unique No Visual Unique With Visual Evasive Conflicts 1 1 2 0 1 VISUAL IMPARED FIELD SURVEY The visually impaired field survey involved a select group of visually impaired and legally blind individuals at the test site. Each participant was directed through each test scenario to test their understanding of the existing and modified audible devices. The results were assessed by a combination of behavioral measures, and their reactions to each of the devices. The Korve team coordinated with Sherry Repscher, the Utah Transit Authority (UTA) ADA Compliance Officer. Sherry works closely with the Utah Council of the Blind and State Services for the Blind and arranged for people to participate in the field test. UTA provided transit service to get participants to the test site if they needed assistance with transportation. The visual impaired field survey consisted of 21 volunteers who crossed the 50 South Main Street crossing when a train was present and then responded to a survey. Among the volunteers tested were10 males and 11 females. Of these individuals, one person was Hispanic and 20 were Caucasian. There were seven individuals that used a guide dog and 20 individuals that received training in the use of the long cane. The training in the use in the long cane varied from one week to through out school. The results of the visual impaired field survey come from an analysis of the responses to certain research questions. Research Question 1: Is there any difference between the reported acceptability of the signals from the devices? Responses to the question, “How acceptable was the signal?” were assigned numbers: 1=Poor, 2=Fair, 3=OK, 4=Very Good, and 5=Excellent. The mean responses for the devices were Bell= 3.52, Baseline=3.62, and Unique=2.48 indicating that the Unique signal was rated as less acceptable than the other two. Research Question 2: Is there any difference between the reported aversive or annoying nature of the signals from the devices? Responses to the question “How aversive or annoying was the signal?” were assigned numbers: 1=Not at all, 2=Very Little, 3=OK, 4=Unpleasant, 5=Nasty. The mean responses for the devices were Bell=2.00, Baseline=1.29, and Unique=2.67, indicating that the Baseline condition was considered the least annoying of the signals and the Unique signal was considered the most annoying. Research Question 3: Is there any difference between the reported usefulness or effectiveness of the signals from the devices? In response to the question “Overall, how useful or effective was the signal as a warning?” participants were asked to rate the

Field Test Report usefulness from 1 to 7 with 1 meaning “not at all” and 7 “very useful”. The analysis showed no significant difference between the signals. Research Question 4: Which device was considered “best” by the subjects? Bell 9 votes 42.9% Baseline 8 votes 38.1% Unique 4 votes 19.0% Research Question 5: Which device was considered “second” by the subjects? Bell 11 votes 52.4% Baseline 5 votes 23.8% Unique 5 votes 23.8% Research Question 6: Which device was considered “worst” by the subjects? Bell 1 vote 4.8% Baseline 8 votes 38.1% Unique 12 votes 57.1% Research Question 7: Is dog guide or cane use related to device considered best? Bell Baseline Unique Dog Guide 5 6 3 Cane 4 2 1 Analyses indicated no significant difference responses of the dog guide and cane users. Research Question 8: Is there a relationship between hearing loss, use of hearing aids, duration of vision loss, and acuity with the rating of device specific acceptability? None of these factors were found to be significantly related to device specific acceptability. Research Question 9: Is there a relationship between subjects who travel often and travel independently with the rating of device specific acceptability? Neither factor was found to be significantly related to device specific acceptability. Research Question 10: Is there a relationship between observer rating of the subjects cane skills with the rating of device specific acceptability? No significant relationship was found between cane skills and ratings of device specific acceptability. In summary, the Unique sound was found to be the least acceptable and the most aversive. The Bell sound was voted “best” and “second” with the Unique sound voted “worst”. There were no significant relationships found between participant responses and several differences among participants, (i.e. cane vs. guide dog users and travel often and independently vs. otherwise).

Field Test Report PUBLIC SURVEYS Surveys of the general public were completed on days when the crossing was being video taped at street level for the behavior observation. When the surveyor heard the audible signal, he/she would look for any pedestrians waiting to cross that were definitely hearing the signal. After the pedestrians were given a walk signal and had crossed the street, he/she would attempt to stop one of the persons noted earlier and ask them the questions on the survey sheet. One intention of the survey process was to question a diverse group of people in regards to age, gender, and frequency of using the crossing. While the sample size (around 30 surveys per scenario) is too small to make any conclusions about the opinions of specific demographics, this approach helped insure that a broad range of views were heard. The results of the survey come from 106 participants. The participants evaluated one of the following scenarios (number of surveys for each scenario is also provided): • Bell Sound – No Visual (24 surveys) • Bell Sound – With Visual (30 surveys) • Unique Sound – No Visual (28 surveys) • Unique Sound – With Visual (24 surveys) The participants included of 52 males and 54 females. There is a good distribution of ages among the participants with most between 20 and 40 years of age. About 85% of the participants use UTA. However, the frequency varies from once a month to more than four times per day with most (38%) of the participants using UTA 2-4 times per day. The audible signal was noticed by at least 83% of the participants in each scenario. There was a diverse understanding of how to respond reported by the participants. Table 8 shows the percent of each response in each scenario. The responses were divided into two groups; correct responses and incorrect responses. The correct responses include “caution”, “look around”, “look for train”, “stop and wait”, and “stop, but was confused at first”. Incorrect responses include “don’t know”, “confused”, “cross anyway”, and “ignore and rely on ped signal”. Other and blank refer to responses that did not directly answer the question and were unanswered respectively. The correct and incorrect response group percentages are also included in Table 8. Throughout the system the train gong is only rung just before a train departs the station or when a train is arriving and a pedestrian is on or near the tracks. Prolonged sounding of a crossing bell would not be a typical stimulus at urban intersections with traffic signal control. Many respondents also associated bell alone with look around, and Unique sound with look for train and to a lesser extent the bell with and without visual with look for train. Stop and wait was associated with stop and wait only when the visual was present. In general correct responses were associated more highly with the unique sound with no visual train warning display and the bell with a visual display, which appears to be somewhat contradictory. Incorrect responses were more associated with the bell with or without the visual display. Given the small sample size for this survey it is difficult to draw any significant conclusions. However, the data tends to indicate three points worth noting:

Field Test Report • No particular treatment stands out as the best (even though the unique audible with visual icon had the fewest incorrect responses, the proportion of correct responses was lower than some of the competing treatments); • The response to the standard crossing bell sound was more consistent with or without the visual icon display; and • The unique audible was more comprehensible when used with the visual icon. Table 9 Survey Responses to the Audible Signal Responses Unique No Visual Unique With Visual Bell No Visual Bell With Visual Caution 0% 0% 0% 0% Look around 0% 13% 25% 10% Look for train 50% 8% 29% 27% Stop and wait 7% 25% 0% 20% Stop, but was confused at first 0% 0% 0% 0% Correct Response Total 57% 46% 54% 57% Don't know 14% 0% 0% 10% Confused 0% 0% 21% 13% Cross anyway 0% 0% 4% 3% Ignore and rely on ped signal 0% 0% 4% 0% Incorrect Response Total 14% 0% 29% 23% Other 25% 38% 13% 7% Blank 4% 17% 4% 10% Correct Responses Excluding Other & Blank Responses 80% 100% 65% 71% The following paragraphs describe the results of four additional research questions that were asked of the participants. Research Question 1: Is there any difference between the reported acceptability of the signals from the devices? The responses to the question, “How acceptable was the signal?” were assigned numbers: 1=Poor, 2=Fair, 3=OK, 4=Very Good, 5=Excellent. The mean responses for the devices were Unique No Visual=3.9, Unique With Visual=3.4, Bell No Visual=3.5, and Bell With Visual=3.6 indicating that there is not much difference between unique scenarios and the bell scenarios. Research Question 2: Is there any difference between the reported aversive or annoying nature of the signals from the devices? The responses to the question “How aversive or annoying was the signal?” were also assigned numbers: Not at all=1, Very Little=2, OK=3, Unpleasant=4, and Nasty=5. The mean responses for the devices were Unique No Visual=2.4, Unique With Visual=2.6, Bell No Visual=2.0, and Bell With Visual=2.4 indicating that there was not much difference in the aversive nature of the unique and bell scenarios. Research Question 3: Is there any difference between the reported usefulness or effectiveness of the audible signals from the devices? The responses to the question

Field Test Report “Overall, how useful or effective was the audible signal as a warning?” asked participants to rate the audible signal on a scale of 1 to 7 with 1 meaning “not at all” and 7 meaning ”very useful”. The mean responses for the devices were Unique No Visual=5.6, Unique With Visual=4.7, Bell No Visual=5.5, and Bell With Visual=5.3. This indicates that the there is not much difference between viewed effectiveness of the bell audible and the unique audible signal. Research Question 4: Is there any difference between the reported usefulness or effectiveness of the visual signals from the devices? The responses to the question “Overall, how useful or effective was the visual signal as a warning?” asked participants to rate the visual signal on a scale of 1 to 7 with 1 being not very not useful and 7 being very useful. The mean responses for the devices with a visual signal were Unique With Visual=4.3 and Bell With Visual=4.8. Limited conclusions can be made here due to the low number of participants in this question for the both scenarios. In summary, the two scenarios with the most participants responding incorrectly to how to respond to the signal were the Bell With no Visual and the Bell With Visual. This indicates that the unique sound may better communicate to pedestrians the proper response. The results of the other research questions show little difference between the four scenarios. It should be remembered that the unique sound was novel and the effect could be a novelty effect. The sound would need to be in place for a much longer period of time before firm conclusions could be drawn. CONCLUSIONS The overall results of the field study indicate that the addition of audible warnings or audible warning with a visual warning display had little effect on the percentage of persons that complied with the pedestrian signal when a train was approaching or departing the station. However the use of the bell was associated with more pedestrian violations, more pedestrians dashing across the rails when violating and fewer pedestrians violating the signal 10 seconds or less prior to the train arrival at the crossing location. These mixed data need to be viewed in regard to the audible warnings already in place during baseline at these sites. It is standard UTA policy to install audible accessible pedestrian signals at level grade rail crossings to reduce the risk of crashes between blind travelers and motor vehicles and light rail trains. These signals may also benefit sighted individuals. It is also policy for the motorman to sound the train’s gong twice as the train departs the station and to sound the gong when arriving at the station when pedestrians are in the train’s path or near the tracks. These procedures provide a salient audible warning in close proximity to the train traversing the crossing. It is possible that these warnings may have produced a ceiling effect at UTA sites that washed out the effect of the added warnings. The addition of the bell providing a earlier warning of the train approach may have motivated some travelers to dash to clear the train with a wider safety margin which resulted in a smaller percentage of pedestrians crossing 10 seconds or less prior to the train reaching their crossing location. The bell may have produced a larger effect

Field Test Report because it is a traditional auditory warning associated with the approach of a train, particularly at high-speed rail crossings. The visual impaired field survey indicated that the Bell sound was the most desirable. However, the accessible pedestrian signal in place during baseline provided a better warning because it indicated to blind pedestrians when it was safe to cross. It is unclear how a train warning could add to the effectiveness of the accessible signal because blind pedestrians would be unlikely to cross in the absence of the accessible signal because it is highly likely they would be struck by a fast moving motor vehicle while crossing the roadway even in the absence of the much slower moving trains. The public survey showed little difference between the four scenarios or the two sounds. However, fewer participants responded incorrectly to the two test scenarios with the Unique audible warning. This indicates that the unique sound may better communicate to pedestrians the proper response. The results of the other research questions show little difference between the four scenarios. It should be remembered that the unique sound was novel and the difference obtained could be a novelty effect. The sound would need to be in place for a much longer period of time before firm conclusions could be drawn. Based solely on the results of the field test of this application we do not recommend a change in current practices at railway crossings. These data are also in accord with the results of the review of the crash data that indicate that light rail crossings are relatively safe. Crashes between pedestrians and LRT are few in number but when crashes occur they are severe; therefore, it is critical to do everything possible to improve safety. Although pedestrians violating the signal might be somewhat less likely to cross just before the train arrives when the bell is present, it is also the case that more pedestrians can be expected to violate the signal when the bell is present. Because of the low incidence of evasive conflicts that served as the crash surrogate measure in this study, it is unclear how these two effects would interact with crashes. The results are not compelling; however, there is some promise for the unique sound. If there is interest in further pursuing the “blended staircase” sound the following things should be considered: • Testing over an extended period of time, • Testing in a wider range of environments, and • Investigating the use of a speaker that would provide a higher fidelity. Further testing would be appropriate for other “alternative treatments” of pedestrian crossings including but not limited to changes in visual and audible devices.