Guidelines for the Application of Crossing Solutions at Roundabouts and Channelized Turn Lanes for Pedestrians with Vision Disabilities (2016)

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

NCHRP 3-78b: Final Project Report April 2016 7 APPENDIX A: WAYFINDING DATA DETAILS This appendix presents the wayfinding method and the data collected at roundabouts and channelized turn lanes. Data were collected at roundabouts in Michigan, North Carolina, Ohio, and Maryland, and at channelized turn lanes in Colorado, North Carolina, Maryland, and Arizona. The data are presented, first, in the form of summary figures and, later, in the form of tables showing the results for individual wayfinding tasks at individual sites. The information is presented in this way in order to assist the reader to understand the general patterns obtained and the substantial variability in wayfinding performance across sites, and to allow the reader to understand the limitations of the research approach and related data. 7.1 Introduction As noted in Section 4, above, the overall crossing task for blind pedestrians at roundabouts and channelized turn lanes consists of four major subtasks: 1. Determining the crossing location, including detecting the street. 2. Aligning to cross (i.e., establishing a correct initial heading) at a crosswalk that may or may not be aligned perpendicular to the sidewalk or in the same direction as the slope of the associated curb ramp. 3. Deciding when to cross in an environment of largely uninterrupted traffic flow, requiring the identification of appropriate gaps in traffic or crossing opportunities in front of yielding vehicles. When signals are provided, an audible message conveys to a blind pedestrian when the WALK sign is active. 4. Maintaining correct heading (staying within the crosswalk) during the entire crossing (e.g., in the case of a roundabout, until the splitter island or far curb is reached). Three aspects of the crossing task are typically characterized as wayfinding tasks: determining the crossing location (finding the crosswalk), aligning to cross, and maintaining crossing heading. Errors in performing these tasks may lead to beginning the crossing outside the crosswalk area, crossing to the central island of a roundabout, or missing the island at a channelized turn lane. This can cause general confusion and disorientation for the blind pedestrian. Crossing at a location that is not within the crosswalk and where drivers are not expecting pedestrians can be a safety issue as well. 7.2 Method 7.2.1 Data Collection A wayfinding study was performed at a subset of the project’s roundabout and channelized turn lane crosswalks to determine error rates for the three major wayfinding tasks. We also recorded information, not summarized in this appendix, on the need for safety interventions by the orientation and mobility (O&M) specialist monitoring the trial and the amount of time taken to perform the wayfinding tasks. For each trial, a blind participant was guided by an O&M specialist to starting locations that varied from over 100 feet from the crosswalk to approximately 10 feet from the crosswalk. The participant was then asked to find the crosswalk, to align to cross, and to cross to the island when asked by the O&M specialist. In addition, the participant was asked to find the crossing starting location on the island and align for the second stage of the crossing. A trial ended after the alignment task on the island; the participant was not asked to complete the second stage of the crossing. Separate data were collected for wayfinding to the island and wayfinding on the island. The sites varied widely in terms of characteristics that the team’s O&M specialists hypothesized would 103

NCHRP 3-78b: Final Project Report April 2016 provide useful wayfinding information. Because the wayfinding study was a component of a larger study whose principal goal was to evaluate crosswalk treatments’ effects on the decision about when to cross (subtask 3 above), the choice of wayfinding sites were constrained by the choices made for the larger study. The data-coding scheme is given in Table 7-1. The raw data codes represent what was actually measured while the analysis codes show how the raw data were combined to arrive at the error figures discussed in this appendix. For example, whereas we recorded the right or left direction of any initial alignment error, these leftward and rightward errors were combined to arrive at what this appendix discusses as “Initial Alignment Error.” There were six major error criteria: 1) Missed Crosswalk Error, 2) Crosswalk Location Error, 3) Initial Alignment Error, 4) Excessive Time Error, 5) Crossing Ending Location Error, and 6) Intervention. A missed crosswalk error indicated that the participant continued past the correct crossing location, the crosswalk, or did not walk far enough to find the crosswalk. Often, participants passed the crosswalk and turned around to search for it without any prompting by the O&M specialist. If the participant continued more than 20 feet past the crosswalk, the O&M specialist would intervene and tell the participant that they had passed the crosswalk. That was coded both as an intervention and a missed crosswalk error. For the raw data, Missed Crosswalk Error was coded as 1) “Did not miss the crosswalk,” 2) “Missed the crosswalk, but corrected,” or 3) “Missed the crosswalk.” For the purposes of deriving an error rate, “Missed the crosswalk” was recoded as 1, and codes 1 and 2 were recoded as 0. A crosswalk location error occurred when a participant aligned to cross outside the crosswalk lines. Crosswalk location error was coded as 1) “Within crosswalk,” 2) “Outside crosswalk <5 feet,” and 3) “Outside crosswalk >5 feet” for the raw data. Codes 2 and 3 were recoded as 1, and “Within crosswalk” was recoded as 0 for the analysis. The initial alignment was recorded when the participant determined that they were aligned and ready to cross. The data coder, standing directly behind participants, observed their basic body and foot alignment and coded their alignment as within crosswalk or outside crosswalk presuming that they maintained their initial heading while crossing. If participants who maintained their initial alignment would complete their crossings outside crosswalk, the direction of misalignment, left or right was recorded. Initial Alignment Error was coded as 1) “Within,” 2) “Left,” and 3) “Right” in the raw data. Codes 2 and 3 were recoded as 1, and “Within” was recoded as 0 for the descriptive statistics analysis. After the alignment was coded, the O&M specialist chose a safe time to cross and asked the participant to cross. The participants crossed independently using their initial alignment or crossing heading and typical cane skills. The data coder recorded the location where participants completed their crossings to the island in the same way as the Initial Alignment Error, as 1) “Within,” 2) “Left,” and 3) “Right” for the raw data. Codes 2 and 3 were recoded as 1, and “Within” was recoded as 0 to derive the Crossing Ending Location Error. Excessive Time Error was initially intended as an efficiency measure, and was based on the amount of time in seconds taken for the entire wayfinding process, with a unique time for wayfinding to the island and wayfinding on the island to the second crosswalk starting location. This error is not included in the descriptive statistics because we could sometimes not control the variability in starting location/trial distance across sites. There were several times when the monitoring O&M specialist might intervene during a trial. An intervention might be recorded if the participant passed the crosswalk, as noted before. Interventions also took place if the participant was beginning to cross into traffic at a dangerous place, such as aligning to cross the circulatory roadway toward the center island at a roundabout, or missing the island completely at a channelized turn lane. The O&M specialist did not typically intervene if the participant made an error but was not in danger. For example, some individuals ended their crossing several feet outside of the crosswalk and walked along the curb of the island in the street while looking for the cut-through area or 104

NCHRP 3-78b: Final Project Report April 2016 curb ramp of the island, but vehicles were not close or were yielding until the person reoriented themselves. Intervention was coded as 1 if at least one intervention occurred during a trial, and 0 if no intervention occurred during a trial. For some participants, there were several interventions during a trial. Interventions occurred during 33 of the 243 total trials. The most interventions to occur in a trial was six, and the average number of intervention across all trials where interventions occurred was one. Missing information was coded as 999, which was used, rarely, when a researcher failed to write down observations about a trial. The code 777 was used when a task was not performed due to circumstances such as participant confusion or safety concerns. There were times when the participant became so disoriented that the O&M specialist ended the trial. Some situations where this took place included an island at a channelized turn lane where the participant ended their first crossing outside the crosswalk and cut-through area and was unable to reorient on the island, thinking the cut-through pedestrian channel was the street. If a participant lined up to cross into the circulatory roadway at the roundabout, the O&M specialist would intervene and guide the person to the crosswalk. In that case the crosswalk location error and subsequent intervention was coded as “Missed the crosswalk” because the participant did not find the crosswalk location independently. Error rates (dichotomous summary data) were calculated by dividing the total number of errors by the sum of errors and no errors for a given task. Error rates were summed by task for all roundabouts and for all channelized turn lanes. Error rates were also calculated separately for each roundabout and each channelized turn lanes. Table 7-1 shows the raw data codes entered during the experiment, which includes up to three key codes for each criterion. For analysis, all criteria were converted into a one/zero binary response, as summarizes in the “analysis codes” column. Table 7-1: Wayfinding Data Codes Based on Error Criterion Raw Data Codes Codes 1 Missed Crosswalk Error 1 Did not miss 1 Missed 2 Missed but corrected 0 Did not miss and Missed but corrected independently 3 Missed 2 Crosswalk Location Error 1 Within CW 1 Outside <5' and Outside >5' 2 Outside <5' 0 Within CW 3 Outside >5' 3 Initial Alignment Error 1 Within 1 Left a nd Right 2 Left 0 Within 3 Right 4 Excessive Time Error Raw time in seconds Raw time in seconds 5 Crossing Ending Location Error 1 Within CW 1 Outside <5' and Outside >5' 2 Outside <5' 0 Within 3 Outside >5' 6 Intervention (any number per trial) 1 Yes 1 Yes 2 No 0 No Missing information was coded as 999. The code 777 was used when a task was not performed. 105

NCHRP 3-78b: Final Project Report April 2016 7.3 Results 7.3.1 Wayfinding to Island Figures 7-1, 7-2, and 7-3 provide error rates for wayfinding to the island for all roundabout and channelized turn lane sites, and for roundabout and channelized turn lane sites by location. Table 7-2 provides detailed information on study locations and sample sizes for roundabout and channelized turn lane sites by intersection location. CTL Roundabout Figure 7-1: To Island Wayfinding Error Rates for All Roundabout and Channelized Turn Lane Sites Bar graph shows the wayfinding error rates for crossings to the island for all sites. 4% missed crosswalk error was recorded at CTL sites and 25% missed crosswalk error for roundabout sites. Crosswalk location errors were 8% for CTLs and 2% at roundabouts. The initial alignment errors for were 43% and 34% for CTLs and roundabouts, respectively. Crossing Ending location errors were 38% at CTLs and 18% at roundabouts. Interventions occurred on 4% of trials at CTLs and 21% of trials at roundabouts. 106

NCHRP 3-78b: Final Project Report April 2016 Figure 7-2: To Island Wayfinding Error Rates for Roundabout Sites Bar graph shows the wayfinding error rates for each of the roundabout sites. For missed crosswalk, there was 13% error rate at Ann Arbor, 15% at Cary, 38% at Hilliard, 15% at Greenbelt and 31% at Oakland County. Crosswalk location errors were 3% at Ann Arbor and 3% at Hilliard and 0% at other sites. The initial alignment errors varied from 10% at Ann Arbor to 55% at Cary, with 34% at Columbus- Hilliard, 40% at Greenbelt and 42% at Oakland County. Crossing ending location errors were 3% at Ann Arbor, 5%at Cary, 24% at Columbus-Hilliard, 45% at Greenbelt and 19% at Oakland County. Interventions occurred on 6% of trials at Ann Arbor, 10% of trials at Cary, 38% of trials at Columbus- Hilliard, 38% of trials at Greenbelt and 19% of trials at Oakland County. Figure 7-3: To Island Wayfinding Error Rates for Channelized Turn Lane Sites Bar graph shows the wayfinding error rates for each of the CTL sites. For missed crosswalk, there was 2% error rate in Boulder, 20% at Cary site, 10% at Greenbelt, and 0% in Tucson. Crosswalk location errors were 8% at Boulder and 20% at Cary, 0% at Greenbelt and 4% in Tucson. The initial alignment errors were 43% at Boulder, 60% at Cary, 70% at Greenbelt and 25% at Tucson. Crossing ending location errors were 41% at Boulder, 20% at Cary, 70% at Greenbelt and 25% at Tucson. Interventions occurred on 3% of trials at Boulder, 0% of trials at Cary, 20% of trials at Greenbelt and 0% of trials in Tucson. 107

NCHRP 3-78b: Final Project Report April 2016 Table 7-2: To Island Wayfinding Locations and Sample Sizes Site Name Site Type Intersection Approach Number of Subjects Total Number of Observations for each task at each site Boulder, CO CTL Arapahoe and Foothills SW N/A 5 10 Boulder, CO CTL Foothills and Baseline NE N/A 5 10 Boulder, CO CTL Foothills and Baseline SW N/A 5 10 Boulder, CO CTL Canyon and 28th SW N/A 5 10 Boulder, CO CTL Pearl and 28th NE N/A 6 12 Boulder, CO CTL Pearl and 28th NW N/A 5 10 Cary, NC CTL Kildaire Farm and Tryon W N/A 5 10 Greenbelt, MD CTL Kenilworth and West NW N/A 6 10 Tucson, AZ CTL Grant and Oracle NE N/A 3 (Pilot subject omitted) 6 Tucson, AZ CTL Grant and Oracle SW N/A 3 (Pilot subject omitted) 6 Tucson, AZ CTL Speedway and Wilmot NW N/A 3 (Pilot subject omitted) 6 Tucson, AZ CTL Tanque Verde and Sabino Canyon NE N/A 3 (Pilot subject omitted) 6 Ann Arbor, MI RBT Ellsworth and State W Entry 4 8 Ann Arbor, MI RBT Ellsworth and State W Exit 4 8 Ann Arbor, MI RBT Nixon and Huron S Entry 4 8 Ann Arbor, MI RBT Nixon and Huron S Exit 4 7 Cary, NC RBT Old Apex and Chatham W Entry 5 10 Cary, NC RBT Old Apex and Chatham W Exit 5 10 Columbus- Hilliard, OH RBT Cemetery and Main N Entry 6 17 Columbus- Hilliard, OH RBT Cemetery and Main S Entry 6 22 Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Entry 6 10 Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Exit 6 11 Oakland County, MI RBT Farmington and Maple E Entry 4 8 Oakland County, MI RBT Farmington and Maple E Exit 2 3 Oakland County, MI RBT Farmington and Maple S Entry 5 9 Oakland County, MI RBT Farmington and Maple S Exit 3 6 For wayfinding to the island, initial alignment had the highest error rates across all roundabout sites and all channelized turn lane sites. For roundabout sites, the initial alignment error rate was highest at the Cary, North Carolina location. For channelized turn lane sites, the initial alignment error rate was highest 108

NCHRP 3-78b: Final Project Report April 2016 at the Greenbelt, Maryland location. Participants were more likely to miss the crosswalk, initially align outside the crosswalk, and end their crossing outside of the crosswalk at the channelized turn lane sites than the roundabout sites when wayfinding to the island. However, there were more interventions overall at the roundabout sites than the channelized turn lane sites, with the highest number of interventions occurring at the Hilliard, Ohio location (15 out of 39 total trials had at least one intervention). 7.3.2 Wayfinding on Island Figures 7-4, 7-5, and 7-6 provide error rates for wayfinding on the island for all roundabout and channelized turn lane sites, and for roundabout and channelized turn lane sites by location. Table 7-3 provides detailed information on study locations and sample sizes for roundabout and channelized turn lane sites by intersection location. CTL Figure 7-4: On Island Wayfinding Error Rates for All Roundabout and Channelized Turn Lane Sites Bar graph shows the wayfinding error rates for crossings on the islands for all sites. 25% missed crosswalk error was recorded at CTL sites and 22% missed crosswalk error for roundabout sites. Crosswalk location errors were 8% for CTLs and 5% at roundabouts. The initial alignment errors for were 33% and 23% for CTLs and roundabouts, respectively. Interventions occurred on 10% of trials at CTLs and 17% of trials at roundabouts. 109

NCHRP 3-78b: Final Project Report April 2016 Only one on island trial at Greenbelt location Figure 7-5: On Island Wayfinding Error Rates for Roundabout Sites Bar graph shows the wayfinding error rates on islands for each of the roundabout sites. Note states “Only one on island trial at Greenbelt location.” For missed crosswalk, there was 10% error rate at Ann Arbor, 20% at Cary, 38% at Columbus-Hilliard, 100% at Greenbelt and 8% at Oakland County. Crosswalk location errors were 10% at Ann Arbor and 6% at Cary, 4% at Columbus-Hilliard, 0% at Oakland County and not recorded at Greenbelt. The initial alignment error was 6% at Ann Arbor 22% at Cary, 28% at Columbus-Hilliard, no trial at Greenbelt, and 41% at Oakland County. Interventions occurred on 3% of trials at Ann Arbor, 15% of trials at Cary, 30% of trials at Columbus-Hilliard, 100% of trials at Greenbelt and 13% of trials at Oakland County. Figure 7-6: On Island Wayfinding Error Rates for Channelized Turn Lane Sites Bar graph shows the wayfinding error rates for each of the CTL sites. For missed crosswalk, there was 16% error rate in Boulder, 80% at Cary site, 50% at Greenbelt, and 17% in Tucson. Crosswalk location errors were 9% at Boulder and 0% at Cary, 13% at Greenbelt and 5% in Tucson. The initial alignment errors were 37% at Boulder, 50% at Cary, 13% at Greenbelt and 30% at Tucson. Interventions occurred on 8% of trials at Boulder, 0% of trials at Cary, 50% of trials at Greenbelt and 4% of trials in Tucson. 110

NCHRP 3-78b: Final Project Report April 2016 Table 7-3: On Island Wayfinding Locations and Sample Sizes Site Name Site Type Intersection Approach Number of Subjects Number of Observations Boulder, CO CTL Arapahoe and Foothills SW NA 5 10 Boulder, CO CTL Foothills and Baseline NE NA 5 10 Boulder, CO CTL Foothills and Baseline SW NA 5 10 Boulder, CO CTL Canyon and 28th SW NA 5 10 Boulder, CO CTL Pearl and 28th NE NA 6 12 Boulder, CO CTL Pearl and 28th NW NA 5 10 Cary, NC CTL Kildaire Farm and Tryon W NA 5 10 Greenbelt, MD CTL Kenilworth and West NW NA 6 10 Tucson, AZ CTL Grant and Oracle NE NA 3 (Pilot subject omitted) 6 Tucson, AZ CTL Grant and Oracle SW NA 3 (Pilot subject omitted) 6 Tucson, AZ CTL Speedway and Wilmot NW NA 3 (Pilot subject omitted) 6 Tucson, AZ CTL Tanque Verde and Sabino Canyon NE NA 3 (Pilot subject omitted) 6 Ann Arbor, MI RBT Ellsworth and State W Entry 4 8 Ann Arbor, MI RBT Ellsworth and State W Exit 4 8 Ann Arbor, MI RBT Nixon and Huron S Entry 4 7 Ann Arbor, MI RBT Nixon and Huron S Exit 4 6 Cary, NC RBT Old Apex and Chatham W Entry 5 10 Cary, NC RBT Old Apex and Chatham W Exit 5 10 Columbus- Hilliard, OH RBT Cemetery and Main N Entry 6 15 Columbus- Hilliard, OH RBT Cemetery and Main S Entry 6 22 Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Entry No Data No Data Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Exit 1 1 Oakland County, MI RBT Farmington and Maple E Entry 4 7 Oakland County, MI RBT Farmington and Maple E Exit 2 3 Oakland County, MI RBT Farmington and Maple S Entry 5 9 Oakland County, MI RBT Farmington and Maple S Exit 3 5 Error rate findings for wayfinding on the island are similar to error rate findings for wayfinding to the island. For wayfinding on the island, initial alignment had the highest error rates across all roundabout sites 111

NCHRP 3-78b: Final Project Report April 2016 and all channelized turn lane sites. For roundabout sites, the initial alignment error rate was highest at the Oakland County, Michigan location. For channelized turn lane sites, the initial alignment error rate was highest at the Cary, North Carolina location. On the island, participants were more likely to miss the crosswalk at the channelized turn lane sites than the roundabout sites. This is in stark contrast to our finding that when traveling to the island, CTL crosswalks were likely to be located (see Figure 7-1 and Table 7-5). There were more interventions overall at the roundabout sites than the channelized turn lane sites, with the highest number of interventions occurring at the Columbus-Hilliard, Ohio location (11 out of 37 total trials had at least one intervention). The 100% intervention rate at the Greenbelt, Maryland location is not reliable, since only one trial was recorded for wayfinding on the island, and this trial had an intervention. 7.3.3 Error Rates for Each Intersection Location More detailed information on error rates for each intersection location is provided in Tables 7-4, 7-5, 7-6, and 7-7. Details for error criterion 1 (Missed Crosswalk Error), criterion 3 (Initial Alignment Error), and criterion 5 (Crossing Ending Location Error) are provided for trials to the island, and details for error criterion 1 (Missed Crosswalk Error) and criterion 3 (Initial Alignment Error) are provided for trials on the island. It should be noted that the crossing cues were not consistent even for two crossings at the same intersection, so the data is recorded for each approach or crosswalk. 112

NCHRP 3-78b: Final Project Report April 2016 Table 7-4: To Island Error Rates for Criterion 1, 3, and 5 at Roundabout Sites Site Name Site Type Intersection Approach Missed Crosswalk Success Missed Crosswalk Failure Missed Crosswalk % Failure Initial Alignment Success Initial Alignment Failure Initial Alignment % Failure Crossing Ending Location Success Crossing Ending Location Failure Crossing Ending Location % Failure Ann Arbor, MI RBT Ellsworth and State W Entry 7 1 13% 7 1 13% 7 1 13% Ann Arbor, MI RBT Ellsworth and State W Exit 8 0 0% 6 2 25% 8 0 0% Ann Arbor, MI RBT Nixon and Huron S Entry 8 0 0% 8 0 0% 8 0 0% Ann Arbor, MI RBT Nixon and Huron S Exit 6 1 14% 7 0 0% 7 0 0% Cary, NC RBT Old Apex and Chatham W Entry 8 2 20% 3 7 70% 10 0 0% Cary, NC RBT Old Apex and Chatham W Exit 9 1 10% 6 4 40% 9 1 10% Columbus- Hilliard, OH RBT Cemetery and Main N Entry 9 8 47% 5 8 62% 4 4 50% Columbus- Hilliard, OH RBT Cemetery and Main S Entry 15 7 32% 18 4 18% 14 4 22% Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Entry 7 3 30% 6 3 33% 7 2 22% Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Exit 10 1 9% 6 5 45% 4 7 64% Oakland County, MI RBT Farmington and Maple E Entry 2 6 75% 4 4 50% 7 1 13% Oakland County, MI RBT Farmington and Maple E Exit 2 1 33% 3 0 0% 2 1 33% Oakland County, MI RBT Farmington and Maple S Entry 7 2 22% 7 2 22% 8 1 11% Oakland County, MI RBT Farmington and Maple S Exit 5 1 17% 1 5 83% 4 2 33% 113

NCHRP 3-78b: Final Project Report April 2016 Table 7-5: To Island Error Rates for Criterion 1, 3, and 5 at Channelized Turn Lane Sites Site Name Site Type Intersection Missed Crosswalk Success Missed Crosswalk Failure Missed Crosswalk % Failure Initial Alignment Success Initial Alignment Failure Initial Alignment % Failure Crossing Ending Location Success Crossing Ending Location Failure Crossing Ending Location % Failure Boulder, CO CTL Arapahoe and Foothills SW 9 1 10% 10 0 0% 5 4 44% Boulder, CO CTL Foothills and Baseline NE 10 0 0% 6 4 40% 4 6 60% Boulder, CO CTL Foothills and Baseline SW 10 0 0% 1 9 90% 3 7 70% Boulder, CO CTL Canyon and 28th SW 10 0 0% 7 3 30% 8 2 20% Boulder, CO CTL Pearl and 28th NE 12 0 0% 6 6 50% 8 3 27% Boulder, CO CTL Pearl and 28th NW 10 0 0% 6 4 40% 7 2 22% Cary, NC CTL Kildaire Farm and Tryon W 8 2 20% 4 6 60% 8 2 20% Greenbelt, MD CTL Kenilworth and West NW 9 1 10% 3 7 70% 3 7 70% Tucson, AZ CTL Grant and Oracle NE 6 0 0% 5 1 17% 5 1 17% Tucson, AZ CTL Grant and Oracle SW 6 0 0% 5 1 17% 4 2 33% Tucson, AZ CTL Speedway and Wilmot NW 6 0 0% 4 2 33% 4 2 33% Tucson, AZ CTL Tanque Verde and Sabino Canyon NE 6 0 0% 4 2 33% 5 1 17% 114

NCHRP 3-78b: Final Project Report April 2016 Table 7-6: On Island Error Rates for Criterion 1 and 3 at Roundabout Sites Site Name Site Type Intersection Approach Missed Crosswalk Success Missed Crosswalk Failure Missed Crosswalk % Failure Initial Alignment Success Initial Alignment Failure Initial Alignment % Failure Ann Arbor, MI RBT Ellsworth and State W Entry 6 2 25% 6 0 0 Ann Arbor, MI RBT Ellsworth and State W Exit 7 1 13% 6 1 14% Ann Arbor, MI RBT Nixon and Huron S Entry 7 0 0% 7 0 0% Ann Arbor, MI RBT Nixon and Huron S Exit 6 0 0% 6 0 0% Cary, NC RBT Old Apex and Chatham W Entry 8 2 20% 7 2 22% Cary, NC RBT Old Apex and Chatham W Exit 8 2 20% 7 2 22% Columbus- Hilliard, OH RBT Cemetery and Main N Entry 6 9 60% 7 0 0% Columbus- Hilliard, OH RBT Cemetery and Main S Entry 17 5 23% 12 8 40% Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Entry No Data No Data No Data No Data No Data No Data Greenbelt, MD RBT Cherrywood and Greenbelt Metro W Exit 0 1 100% No Data No Data No Data Oakland County, MI RBT Farmington and Maple E Entry 6 1 14% 2 4 67% Oakland County, MI RBT Farmington and Maple E Exit 3 0 0% 2 1 33% Oakland County, MI RBT Farmington and Maple S Entry 8 1 11% 8 0 0% Oakland County, MI RBT Farmington and Maple S Exit 5 0 0% 1 4 80% 115

NCHRP 3-78b: Final Project Report April 2016 Table 7-7: On Island Error Rates for Criterion 1 and 3 at Channelized Turn Lane Sites Site Name Site Type Intersection Missed Crosswalk Success Missed Crosswalk Failure Missed Crosswalk % Failure Initial Alignment Success Initial Alignment Failure Initial Alignment % Failure Boulder, CO CTL Arapahoe and Foothills SW 9 1 10% 9 0 0% Boulder, CO CTL Foothills and Baseline NE 7 3 30% 1 8 89% Boulder, CO CTL Foothills and Baseline SW 9 1 10% 5 4 44% Boulder, CO CTL Canyon and 28th SW 10 0 0% 7 3 30% Boulder, CO CTL Pearl and 28th NE 10 2 17% 7 3 30% Boulder, CO CTL Pearl and 28th NW 7 3 30% 5 2 29% Cary, NC CTL Kildaire Farm and Tryon W 2 8 80% 1 1 50% Greenbelt, MD CTL Kenilworth and West NW 5 5 50% 7 1 13% Tucson, AZ CTL Grant and Oracle NE 6 0 0% 5 1 17% Tucson, AZ CTL Grant and Oracle SW 5 1 17% 2 3 60% Tucson, AZ CTL Speedway and Wilmot NW 4 2 33% 3 1 25% Tucson, AZ CTL Tanque Verde and Sabino Canyon NE 5 1 17% 4 1 20% 116