Exploratory Advanced Research Program

VASTO - Evolutionary Agent System for Transportation Outlook

Agent-Based Modeling and Simulation in The Dilemma Zone

4 DZ Data Collection Using a Highway Driving Simulator

4.1 Method

4.1.1 Experimental Design

There were five design factors in this study: (1) facility speed limit, (2) degree of driving in a hurry, (3) presence of a red-light photo enforcement camera, (4) presence of a pedestrian countdown signal, and (5) presence and behavior of an adjacent vehicle. Each design element is discussed in more detail below. For DZ intersections, the facility speed limit and degree of driving in a hurry were between-group factors, while the presence of a red-light photo enforcement camera, presence of a pedestrian countdown signal, and behavior of an adjacent vehicle were within-group factors.

• Facility speed of the roadway (FacilitySpeed): The two speed limits were 40 mi/h (64.37 km/h) and 55 mi/h (88.51 km/h). All drives completed by a participant had the same posted speed limit.
  
  
• Degree of driving in a hurry (InAHurry): Half of the participants were tasked to drive in a hurry (as described in section 4.1.6). However, there was no change in the simulation between those tasked with driving in a hurry and those not tasked with driving in a hurry. All drives completed by a participant had the same degree of hurried driving.
  
  
• Presence of a red-light photo enforcement camera at the intersection (RedLightCam): Red-light photo enforcement cameras were simulated at four of the DZ intersections and eight of the non-DZ intersections. The intersections with an enforcement camera were appropriately signed (one 500 feet in advance of the intersection and one at the intersection).
  
  
• Presence of a pedestrian countdown signal at the intersection (PedCountSig): Pedestrian countdown signals were simulated at four of the DZ intersections and eight of the non-DZ intersections. Three to five simulated pedestrians crossed or prepared to cross in each direction on the near crosswalk at both of the forced red phase intersections and two of the forced green phase intersections. Pedestrians were included to make the simulation appear more realistic to the driver and to mimic the possible threat of pedestrians at the DZ intersections.
  
  
• Behavior of an adjacent vehicle as the participant approaches the intersection (AdjVehBeh): An adjacent vehicle in the left lane (participant was in the right lane) was present for all DZ intersections and eight of the non-DZ intersections. The vehicle first turned onto the simulated roadway from a left midblock side entrance. It accelerated to overtake the participant's vehicle and maintained a one-car-length lead on the participant's vehicle until the next signal.
  
  

  The vehicle stopped at four of the DZ intersections and proceeded through the other four. If the vehicle was supposed to stop, then it slowed, and the brake lights were simulated 1 second after the onset of the yellow phase. The adjacent vehicle never changed lanes and, in all cases, would turn off at the next midblock location following the green phase of the intersection. In addition, for non-DZ intersections, the presence of an adjacent vehicle was manipulated to expose participants to similar road conditions between DZ and non-DZ intersections. Thus, participants would not know whether they are at DZ intersections or not.

In addition to the aforementioned five factors, oncoming left-turn traffic was considered at eight of the non-DZ green phase intersections to make the simulation appear more realistic to the driver and to mimic the possible threat of a left-turning vehicle at the DZ intersections (in fact, there was no left-turning vehicle at the DZ intersections). Three or four vehicles were queued in the turn lane, with the first vehicle 328.08 feet (100.00 m) away from the intersection. The first vehicle started to move when the participant was 21 seconds away from the intersection. Each successive vehicle started to travel through the intersection 12 seconds after the preceding one. Oncoming traffic traveled, on average, at a rate of 40 mi/h (64.37 km/h). Depending on how fast the participant drove, all, some, or none of these other vehicles passed through the intersection. The oncoming traffic turned in the presence of the driver, but did not interact with the driver in any way.

There are four groups specified by the between‑group factors: (1) facility speed limit, and (2) degree of driving in a hurry. ⁽⁹⁾ The posted speed limit for groups A and B was 40 mi/h, and the posted speed limit for groups C and D was 55 mi/h. Participants in groups A and C drove as they normally would, and those in groups B and D drove in a hurry. A payoff matrix was used to motivate participants to drive in a hurry (see section 4.1.6). The manipulation of the within‑group factors is discussed in more detail in section 4.1.2.

4.1.2 Simulation

The simulation consisted of a four-lane roadway (two lanes in each direction) with signalized intersections present every 0.6 mi (0.97 km). There was also a left turn lane for each direction of travel. Each lane was 12 feet (3.66 meters) wide. The total intersection width (from stop bar to stop bar, including crosswalks, shoulders, curbs, and sidewalks) was 193.55 feet (58.99 meters). The aforementioned configuration of the simulated intersection is shown in Figure 4-1, where from each direction the short white line represents the stop bar and the two longer white lines delineate the two sides of the crosswalk. The total mileage was approximately 14.5 mi (23.34 km) and the simulated environment resembled a suburban commercial area. Assuming participants would comply with the speed limit, for those participants who had a 40-mile-per-hour speed limit, the drive was approximately 22 minutes long without traffic signals and signal compliance (i.e., no stopping while driving). For those participants with a 55-mile-per-hour speed limit, the drive was approximately 16 minutes long, without traffic signals and signal compliance.

© AAI
Figure 4-1: Image. Intersection indicating where pedestrians crossed.

The posted speed limit follows the facility speed limit based on the group assignment, as stated in section 4.1.1 (i.e., 40 mi/h or 55 mi/h). Speed limit signs were posted once every three intersections in the simulation.

To make all participants maintain the posted speed limit, the driver of a vehicle going 8 mi/h over the speed limit received a speeding ticket ($0.50) after that participant's drive (see section 4.1.6). There were 24 intersections in each drive. Eight of the intersections presented a DZ, and the remaining 16 intersections did not. Of the 16 non-DZ intersections, 2 were set to the red phase for all participants and the rest remained green throughout. The order of the 24 intersections within a drive was pseudorandom, and there were 4 different scenario orders (order I-order IV).

Table 4-1, ⁽⁹⁾ Table 4-2, ⁽⁹⁾ Table 4-3, ⁽⁹⁾ and Table 4-4 ⁽⁹⁾ outline order I, order II, order III, and order IV, respectively. One of the orders was randomly selected for each participant and the experiment was conducted according to the selected order. Therefore, the order of the intersections was different for each participant, but for any one participant, the same order was used for his or her drives. Within each of the groups (A-D), 6 of the 24 participants received each order (I, II, III, or IV). The same four scenario orders were replicated with the two speed limits (40 mi/h and 55 mi/h) for a total of eight different simulation designs.

Table 4-1: Order I description by scenario ID, signal phase, and values for independent factors. ⁽⁹⁾

Table 4-2: Order II description by scenario ID, signal phase, and values for independent factors. ⁽⁹⁾

Table 4-3: Order III description by scenario ID, signal phase, and values of independent factors. ⁽⁹⁾

Table 4-4: Order IV description by scenario ID, signal phase, and values of independent factors. ⁽⁹⁾

The distance at which DZs were triggered depended on the facility speed limit. For the 40 mi/h (64.37 km/h) case, the DZ triggered when the driver was 852.30 feet (259.78 m) from the stop bar, and the signal turned yellow 10 seconds later. The yellow phase lasted 4 seconds, and the red phase lasted 38 seconds. For the 55 mi/h (88.51 km/h) case, the DZ triggered when the driver was 812.20 feet (247.56 m) from the stop bar, and the signal turned yellow 5 seconds later. The yellow phase lasted 5 seconds, and the red phase lasted 15 seconds.

The duration of the yellow phase in both cases met the traffic signal control guideline (i.e., yellow change interval) established in the Traffic Engineering Handbook. ⁽¹³⁾ The period of time from when the DZ was triggered until the onset of the yellow phase incorporated the pedestrian countdown signal timing.

The pedestrian countdown started at 9 seconds to the stop bar for the 40 mi/h (64.37) drive and 4 seconds to the stop bar for the 55 mi/h (88.51 km/h) drive.

Oncoming left-turn traffic was present for half (8) of the non-DZ intersections; there was no oncoming traffic at DZ intersections (see Table 4-1 for a more detailed description). There was also no oncoming traffic at the forced red phase intersections. For forced green phase intersections, left-turning vehicles crossed through the intersection before the driver entered the intersection. These vehicles were included to make the simulation appear more realistic to the driver and to mimic the possible threat of a left-turning vehicle at the DZ intersections. The oncoming traffic turned in the presence of the driver but did not interact with the driver in any way. There were three to five pedestrians crossing in each direction on the near crosswalk (see Figure 4-1) at both of the forced red phase intersections and two of the forced green phase intersections; there were no pedestrians at DZ intersections. Pedestrians were included to make the simulation appear more realistic to the driver.

Regardless of whether a DZ was present, the designs of the red-light photo enforcement camera and pedestrian countdown signal were the same. The intersections with a red-light photo enforcement camera were appropriately signed; the signal warning sign (W3-3) ⁽¹⁴⁾ mounting the photo-enforced plaque (W16-10P) ⁽¹⁴⁾ was displayed in advance of the intersection (i.e., 500 feet), and the red-light photo enforced sign (R10-19-DE) ⁽¹⁴⁾ was displayed at the intersection. The light-emitting diodes (LED) pedestrian countdown signal ⁽¹⁴⁾ was located at the intersection.

4.1.3 Highway Driving Simulator

The HDS consisted of a late model compact car chassis mounted inside a section of a cylindrical projection screen. The car was instrumented so that normal driver functions, visual scenes, and audio were interactive.

Figure 4-2: Photograph. Driving simulator used for the study.

The cylindrical screen had a horizontal radius of 107 inches (271.78 cm) and a vertical height of 95 inches (241.30 cm). It wrapped around the center (also known as the eye point) for a total of 240 degrees. When looking at the center of the projections, the driver's eye point was between 0 and 8 inches (20.32 cm) to the rear from the center of the cylinder. Three BARCO SIM10 liquid crystal on silicon digital projectors, each with a resolution of 4,096 by 2,400 pixels, provided the screen image. The luminance of each projector was measured three times. Across the three projectors, the mean luminance was 41.62 foot-lambert (fL); (142.60 candela per square meter (cd/m²), with a range from 40.86 fL (140.00 cd/m²) to 42.26 fL (144.79 cd/m²). The combined image of the 3 projectors covered 200 degrees around the center of the screen. A cluster of personal computers with high-resolution graphics cards were used for rendering the scene image and controlling the real-time scenario. The system updated at a rate of 60 Hertz. The setup of the HDS is shown in Figure 4-2.

4.1.3.1 Post-Participation Questionnaire

Participants completed a questionnaire after their final drive. The questionnaire was administered through MediaLab^® software on a laptop computer. On the questionnaire, participants answered questions regarding whether they stopped at or proceeded through the DZ intersections and why, and whether they noticed or were influenced by the red-light photo enforcement cameras, pedestrian countdown signals, and adjacent vehicles. For the purposes of the questionnaire, DZ intersections were defined as follows: "During each of the simulated drives, some of the traffic signals turned to yellow as you were approaching the intersection. Consider your behavior at these intersections (whether you stopped at or proceeded through the intersection) for the remaining questions." Followup questions regarding red-light running behaviors in the real world were also included. The specific wording for each question is in Appendix A. Post-Participation Survey.

4.1.4 Participants

Originally, 24 participants for each group (A-D) were planned to be recruited in this study. However, considering a 25 percent dropout rate due to simulator sickness, 121 participants were recruited and began testing. Participants were between 18 and 72 years of age, with roughly equal numbers of younger (44 years or younger) and older (45 years or older) drivers in each group. Additionally, there were roughly equal numbers of male and female drivers in each group. All participants had a valid driver's license. Based on the possession of a license, visual acuity was assumed to be at least 20/40 (corrected, if necessary) in at least one eye. Of 121 participants, 19 drivers chose to discontinue due to simulator sickness. Two drivers were forced to discontinue due to problems with the simulator, and one driver was removed from the study due to erratic driving behaviors and rule breaking. The remaining 99 participants (51 males) completed the study. Therefore, responses of the 99 participants were used in this study.

The HDS was also equipped with a six-degree-of-freedom (tilt, roll, and yaw) motion base. The motion base was turned on for approximately the first month of data collection and then turned off to mitigate high rates of simulator sickness. Of those who completed the study, 29 participants drove the study with the motion base turned on, and 70 participants drove the study with the motion base turned off. Participants were also split by age group, either younger (44 years or younger) or older (45 years or older). Approximately half of the participants fell into each age group. The mean age of the males was 47 years (range 19 to 82 years), and the mean age of the females was 46 years (range 18 to 80 years).

4.1.5 Procedures

Each participant completed a practice drive to become acquainted with the simulator. The practice drive includes four intersections with two green and two red signals. The practice roadway mirrored what was used in the test drives, but no manipulations (e.g., pedestrian countdown signals and red-light photo enforcement cameras) were present. No dilemma zones were introduced in the practice drive. If participants had become familiar with the yellow phase timing during the practice drive, they could have made decisions (i.e., accelerate or decelerate their vehicles) before a traffic signal turned to the yellow phase in order to avoid dilemma zones. The signals in the practice drive mimicked the time and duration of the signals used in the test drives. Each participant was instructed to proceed through two green signals and stop at two red signals. If the researcher believed the participant was having trouble with controlling the vehicle on the driving simulator, then the researcher asked the participant to continue until a higher level of comfort was achieved.

Next, the participants completed a series of three drives. Each drive (a series of 24 intersections) lasted approximately 16 to 22 minutes, depending on the posted facility speed (40 mi/h or 55 mi/h) and the number of times the driver stopped at the DZ intersections. Participants took a short break between drives and were asked to complete a simulator sickness questionnaire so that symptoms of simulator sickness could be monitored. ⁽¹⁵⁾ After the third drive, researchers administered the post-participation questionnaire.

4.1.6 Instructions

Researchers verbally informed participants of the instructions before each drive, including before the practice drive. All participants were reminded of the posted speed limit and were asked to stay in the right lane and not change lanes. Participants were offered a $10 bonus, as described below, in addition to the $80 they received for participation. ⁽¹⁶⁾

To advise participants of aerial speed enforcement and red-light photo enforcement in the scenarios, researchers read the following to the participants:

• "You have been told by a friend that this road has aerial speed enforcement. Your friend received a ticket for going 8 miles per hour over the speed limit. If you receive a speeding ticket, you will be charged a fee of $1 from your overall bonus amount.

  There will also be red-light photo enforcement at some of the intersections. The red-light photo enforcement camera is triggered if your vehicle enters the intersection after the signal has turned to red. If you commit a red-light violation at these intersections, then you will lose $0.50 per violation from your overall bonus amount."

If the drivers were to be in a hurry, then researchers also read the following to advise them of penalties if they did not meet their expected completion time (18.5 minutes for the 55 mi/h (88.51 km/h) group and 22.5 minutes for the 40 mi/h (64.37 km/h) group).

• "You are running late for a doctor's appointment. If you finish the simulation in __ minutes, then you will arrive on time for the appointment. However, if you arrive late for the appointment, then you will be charged a late fee of $1 from your overall bonus amount."

As listed above, deductions occurred if drivers were late or speeding. For the speeding deduction in the 40 mi/h (64.37 km/h) case, participants needed to drive 45 mi/h (72.42 km/h) or more for at least 30 seconds. For the speeding deduction in the 50 mi/h (88.51 km/h) case, participants needed to drive 60 mi/h (96.56 km/h) or more for at least 20 seconds.

In addition, drivers received a $1 deduction if they changed lanes. Participants were not penalized beyond the $10 bonus they were originally offered.

4.2 Results

4.2.1 Influence of Motion Base

One concern was whether it was appropriate to combine data from participants who completed the study with the motion base turned on with those who did not. It was believed that this particular study would have a high simulator sickness rate because of the many stops. Given that the inclusion of the motion may have increased the severity and frequency of simulator sickness, one may hypothesize that drivers would, over time, start to proceed through the yellow signal phase to avoid stopping. In other words, the probability of proceeding through a DZ intersection would increase across the three drives.

Generalized estimating equations (GEE) with a binomial response distribution and logit link function were used to determine whether the probability of proceeding through a DZ intersection changed over time for those drivers who completed the study with the motion base turned on and those who completed it with the motion base turned off. The predictors in the model were motion, drive number, and the interaction between the two. Wald Type III statistics indicated that the drive number was the only significant effect with 95 percent confidence interval (CI); χ² (2) = 9.69, p = 0.0078. Both motion and the interaction of motion with drive number were not significant. Therefore, it was deemed appropriate to combine the two groups.

4.2.2 Driver Performance Measures

GEEs with a binomial response distribution and logit link function were used to determine whether the probability of proceeding through a DZ intersection varied regarding the different design factors. An exchangeable correlation structure was assumed. Due to the limited number of observations from 99 participants, only second-order interaction effects were considered. The drive number, which significantly affected the probability of a driver proceeding through a DZ intersection in the previous analysis, was selected as the interaction factor with other factors. Thus, the predictors in the model were facility speed limit (FacilitySpeed), degree of driving in a hurry (InAHurry), presence of a red-light photo enforcement camera (RedLightCam), presence of a pedestrian countdown signal (PedCountSig), behavior of the adjacent vehicle (AdjVehBeh), drive number (Drive), and all second-order interactions with the drive number. Wald Type III statistics for this analysis are shown in Table 4-5. Significant effects with 95 percent CI, printed in boldface red, are discussed in more detail in the following sections.

Table 4-5: Wald Type III statistics for GEE analysis of the probability of proceeding through a DZ intersection regarding the different design factors.

4.2.2.1 Between-Group Factors

There were two between-group design factors: facility speed limit (FacilitySpeed) and degree of driving in a hurry (InAHurry). The main effects for both were significant, and the interaction of the drive number with degree of driving in a hurry was also significant.

Participants in the 40 mi/h (64.37 km/h) speed condition had a higher probability of continuing through a DZ intersection (M = 0.43 with a 95 percent CI of [0.31, 0.56]) than those in the 55 mi/h (88.51 km/h) speed condition (M = 0.25 with a 95 percent CI of [0.17, 0.36]). There are at least two possible rationales for this. First, participants most likely knew from their real-world driving experience that it would take longer to come to a complete stop when traveling at 55 mi/h (88.51 km/h). Thus, these drivers may have anticipated the need to stop and been more conscientious in taking notice and being cautious when the signal changed from green to yellow. Second, based on the developed signal timing (see section 4.1.1), participants in the 55 mi/h (88.51 km/h) speed condition had approximately 0.9 seconds longer to come to a complete stop than their slower counterparts. The time to a complete stop is 3.4 seconds in the 40 mi/h zone and 4.3 seconds in the 55 mi/h zone when the perception reaction time (PRT) is 1 second. Although this was consistent with the physics needed to stop a car traveling at higher speeds, it may have allowed participants more time to come to a complete stop at the DZ intersections; thus, these drivers may have been more willing to stop.

Figure 4-3: Chart. Predicted probability of proceeding through a DZ intersection by degree of driving in a hurry and drive number.

As shown in Figure 4-3, participants who were not tasked to drive in a hurry were fairly consistent in each drive in their probability of proceeding through a DZ intersection (M = 0.11 with 95 percent CI of [0.05, 0.22] at the first drive; M = 0.11 with 95 percent CI of [0.06, 0.20] at the second drive; and M = 0.13 with 95 percent CI of [0.06, 0.23] at the third drive). The bars in Figure 4-3 represent the 95 percent CI of each case. Conversely, participants who were in a hurry increased their probability of continuing through a DZ intersection from the first drive (M = 0.42 with 95 percent CI of [0.29, 0.57]), to the second drive (M = 0.58 with 95 percent CI of [0.44, 0.71]), and again to the third drive (M = 0.63 with 95 percent CI of [0.49, 0.75]). This outcome could perhaps be due to the financial penalty described to participants by the researchers and incorporated into the research procedure to create pressure for the drivers to finish the drive in a certain amount of time. The greatest increase in the probability of proceeding occurred between the first and second drives (a difference of 0.16). Additionally, 94 percent of the in-a-hurry drivers (InAHurry) received a financial penalty after the first drive because they were late for their doctor's appointment. In comparison, only 68 percent of the participants received a financial penalty after the second drive. Between the second and third drives, the probability of proceeding increased by only 0.05. These results imply that the financial penalty described by the researchers succeeded in creating time pressure for the drivers, and being in a hurry significantly increased the probability of a driver proceeding through a DZ intersection. Even on the first drive, participants who were in a hurry were four times more likely to continue through the intersection than their nonpressured counterparts.

4.2.2.2 Within-Group Factors

There were three within-group design factors: presence of a red-light photo enforcement camera (RedLightCam), presence of a pedestrian countdown signal (PedCountSig), and behavior of an adjacent vehicle (AdjVehBeh). All three main effects were significant, and the interaction of the drive number with the presence of an enforcement camera was also significant.

Figure 4-4: Chart. Predicted probability of proceeding through a DZ intersection by presence of a red-light photo enforcement camera and drive number.

As the interaction term with the presence of a red-light photo enforcement camera was significant, the corresponding main effect will not be discussed. The results are displayed in Figure 4-4 where the bars represent the 95 percent CI of each case. For those intersections where an enforcement camera was present, the probability of proceeding increased from the first drive (M = 0.27 with 95 percent CI of [0.17, 0.40]) to the second (M = 0.37 with 95 percent CI of [0.26, 0.48]), but then plateaued and remained constant during the third drive (M = 0.36 with 95 percent CI of [0.26, 0.47]). In comparison, for intersections without a red-light photo enforcement camera, the probability of continuing through a DZ intersection increased across all drives (M = 0.28 with 95 percent CI of [0.18, 0.41] at the first drive; M = 0.35 with 95 percent CI of [0.25, 0.46] at the second drive; and M = 0.42 with 95 percent CI of [0.31, 0.54] at the third drive). During data collection, red-light violations were noted if a participant entered the intersection following the onset of the red phase. However, only three violations occurred among all participants, and the violations were for different people. Drivers were perhaps at first wary of receiving a fine for a red-light violation, but then became more attentive to the placement of the red-light photo enforcement cameras as they became familiar with the simulated scenario.

The probability of proceeding was greater for those intersections without a pedestrian countdown signal (M = 0.41 with 95 percent CI of [0.30, 0.54]) than for those with the countdown signal (M = 0.57 with 95 percent CI of [0.18, 0.38]).

The probability of continuing was greater when the adjacent vehicle also traveled through the intersection (M = 0.36 with 95 percent CI of [0.25, 0.48]) than when it stopped at the signal (M = 0.32 with 95 percent CI of [0.23, 0.44]).

4.2.3 Post-Participation Questionnaire

4.2.3.1 Behavior at DZ Intersections

Two questions were asked to gather information on whether participants stopped at or proceeded through the DZ intersections (see questions 1 and 2 in Appendix A. Post-Participation Survey for more details).

Ninety-eight percent of the participants recalled stopping at least once at a DZ intersection. The reasons participants gave for stopping at the DZ intersections were classified into the following categories: adjacent vehicle related; experiment related (e.g., feeling of being watched); normal driving (e.g., rule-following behaviors); pedestrian related; red-light photo enforcement camera related; and signal timing related. The modal categories were pedestrian related (39 percent) and signal timing (31 percent). Pedestrian-related rationales included the presence of the pedestrian countdown signal (63 percent). Signal-timing rationales included the impending onset of the red phase (33 percent), the distance to the intersection at the onset of the yellow phase (27 percent), the length of the yellow phase (20 percent), and the onset of the yellow phase (20 percent).

Sixty-eight percent of the participants recalled proceeding through at least one DZ intersection. The reasons for continuing through DZ intersections were classified into the following categories: to avoid a potential collision; environmental factors (e.g., pedestrian countdown signal; adjacent vehicle); being in a hurry; being inattentive; confidence in crossing the intersection before the onset of the red phase; and distance to the intersection at the onset of the yellow phase. The modal categories were the distance to the intersection at the onset of the yellow phase (40 percent) and confidence in crossing the intersection before the onset of the red phase (27 percent). Only six percent of the participants cited the pedestrian countdown signal or the adjacent vehicle in their rationales. In addition, approximately 10 percent of the participants provided answers indicating that they may have misunderstood the question.

4.2.3.2 Red-Light Photo Enforcement Camera Influence

Question 3 was asked to gather information on whether participants noticed the red-light photo enforcement camera and, if so, whether the presence of the enforcement camera influenced their behavior (see Appendix A. Post-Participation Survey).

Approximately 75 percent of the drivers claimed to have run a red light in the real world at least once. Of these individuals, the most common reason for doing so was because they could not stop in time.

Fifty-five percent of the participants noticed the red-light photo enforcement camera present at some of the intersections. The following results reflect responses only from these drivers.

Fifty-four percent of the participants believed the presence of the enforcement camera influenced their behavior at the DZ intersections. Additionally, 94 percent of these drivers thought that their behaviors in response to the enforcement camera mimicked what they would have done in the real world. Participants were asked to rate, on a scale from 1 (not at all influenced) to 10 (extremely influenced), how much the red-light photo enforcement camera influenced their decision at the DZ intersections. The mode was 10, and the median rating was 5.5.

4.2.3.3 Pedestrian Countdown Signal Influence

Question 4 was asked to gather information on whether participants noticed the pedestrian countdown signal and, if so, whether the presence of the pedestrian countdown signal influenced their behavior (see Appendix A. Post-Participation Survey).

Ninety-two percent of the participants noticed the pedestrian countdown signal at some of the intersections. The following results reflect responses only from these drivers.

Ninety-three percent of the participants believed the presence of the pedestrian countdown signal influenced their behavior at the DZ intersections. Participants used the remaining time of the pedestrian countdown signal so that they could have a clear decision at an intersection. Additionally, 86 percent of these drivers thought their behaviors in response to the pedestrian countdown signal mimicked what they would have done in the real world. Participants were asked to rate, on a scale from 1 (not at all influenced) to 10 (extremely influenced), how much the pedestrian countdown signal influenced their decision at the DZ intersections. The mode was 10, and the median rating was 8.0.

4.2.3.4 Adjacent Vehicle Influence

Question 5 was asked to gather information on whether participants noticed the adjacent vehicle and, if so, whether the presence and/or behavior of the adjacent vehicle influenced their behavior (see Appendix A. Post-Participation Survey).

All participants noticed the adjacent vehicle traveling in the left lane. Only 13 percent of the drivers believed the presence of the adjacent vehicle influenced their behavior at the DZ intersections. Similarly, only 11 percent of the participants believed the behavior of the adjacent vehicle influenced their behavior at the DZ intersections. Eighty-four percent of the drivers thought their behaviors in response to the adjacent vehicle mimicked what they would have done in the real world.

Participants were asked to rate, on a scale from 1 (not at all influenced) to 10 (extremely influenced), how much the adjacent vehicle influenced their decision at the DZ intersections. The mode was 1, and the median rating was 2.0. However, the results conflicted with the statistical analysis, shown in Table 4-5 (i.e., behavior of an adjacent vehicle (AdjVehBeh) was a significant element to the drivers' responses at the DZ intersection).

4.2.3.5 Real-World Experience with Red-Light Running

Questions 6 to 10 were asked to gather information about participants' self-reported red-light running in actual driving. Respondents were asked to consider only their past experiences in the real world (see Appendix A. Post-Participation Survey).

Approximately 66 percent of participants defined running a red light as entering an intersection on red, while the balance defined running a red light as entering an intersection on yellow and the light turning to red while the car was still in the intersection. Table 4-6 and Table 4-7 show the distribution of how often the participants recalled other drivers and themselves running red lights, respectively.

Table 4-6: Distribution of how often participants recalled other drivers running red lights in the real world (based on their respective definition of red-light running).

Table 4-7: Distribution of how often participants recalled themselves running red lights in the real world (based on their respective definition of red-light running).

Those participants who reported running a red light in the real world were asked to identify which factors affected their decision to do so. Multiple selections were allowed. Figure 4-5 displays what percentage of these drivers selected each factor.

Figure 4-5: Chart. Percentage of participants who selected each factor that affected their decisions to run a red light in the real world.