H. T. Zwahlen, A. Russ, and T. Schnell 1

Transcription

1 H. T. Zwahlen, A. Russ, and T. Schnell 1 Paper No Title: Driver Eye Scanning Behavior While Viewing Ground-Mounted Diagrammatic Guide Signs before Entrance Ramps at Night Helmut T. Zwahlen, Andrew Russ, Ohio Research Institute for Transportation and the Environment 114 Stocker Engineering and Technology Center Ohio University Athens, OH Phone (740) or (740) Fax (740) zwahlen@ohio.edu, ar309001@ohiou.edu And Thomas Schnell Operator Performance Laboratory The University of Iowa Iowa City, IA Phone (319) Fax (319) tschnell@engineering.uiowa.edu Prepared for Presentation at the 82nd Annual Meeting of the Transportation Research Board January 12-16, 2003 National Academy of Sciences Washington, DC November 2002 Number of words: 3656 Number of tables: 0 Number of figures: 10 Total number of words counting tables and figures as 250 words each: 6156

2 2 ABSTRACT The driver eye scanning study reported here is part of a larger study conducted for the Ohio Department of Transportation (ODOT) to evaluate the effectiveness of ground-mounted diagrammatic guide signs placed before entrance ramps at highway freeway interchanges. This nighttime study investigated driver eye-scanning behavior while approaching ground-mounted diagrammatic guide signs placed before entrance ramps. Six highway-freeway interchanges were selected in the Greater Columbus, Ohio, area for the placement of the diagrammatic signs in the field. Six unfamiliar drivers, between the ages of 22-42, were recruited to participate in this study. The diagrammatic sign layout was designed as part of the study in collaboration with ODOT. Two diagrammatic signs were located at each of the six interchanges, ½ mile (805 m) and ¼ mile (402 m) prior to the last point of the gore, where a driver can still gain access to the correct freeway entrance ramp. Driver eye scanning behavior measurements were recorded at night using an ASL4000 measurement system installed in a 1994 Ford Taurus station wagon. The recordings were used to determine if the presence of the diagrammatic signs elicited an excessive number of eye fixations and/or was visually distracting to the drivers. The results of the eye movement analysis indicate that the diagrammatic signs are not looked at excessively often or excessively long. The overall median first look distance to the diagrammatic signs was found to be 125 m. The first sign received an average of 2.9 looks and the second sign received an average of 2 looks, and the average first look durations were 0.74 s and 0.66 s respectively for the first and second signs, and the corresponding last look durations were 0.68 s and 0.62 s. These average look numbers and average look duration times indicate a normal and reasonable level of information acquisition processing employed by the drivers. These values agree with those previously obtained for regular traffic signing determined in previous eye-scanning studies. In summary we conclude that ground mounted diagrammatic signs on multi-lane arterials in advance of highway freeway interchanges do not unduly distract drivers and detrimentally affect a driver s looking behavior.

3 3 BACKGROUND This research was conducted at six interchanges in the greater Columbus, Ohio, area and was a part of a larger study evaluating the effectiveness of ground-mounted diagrammatic guide signs at highway-freeway entrance ramps [1]. This study investigated the nighttime driver eyescanning behavior while approaching ground-mounted diagrammatic guide signs placed ahead of entrance ramps. In this study driver eye scanning behavior was recorded and analyzed during the approaches to the test interchanges. The first look distances, number of looks, duration of the looks, and the last look distances to the diagrammatic guide signs were recorded and analyzed. Ground mounted diagrammatic entrance ramp approach signs, such as the one shown in Figure 1b, are intended to give drivers advance information about the choice of lanes when approaching the entrance ramps at a highway-freeway interchange. The use of ground mounted diagrammatic entrance ramp approach signs will provide a lower cost alternative to the use of overhead span type sign bridges, and represent an improvement over existing trailblazer assemblies. The Federal Manual on Uniform Traffic Control Devices (MUTCD) [2] states that diagrammatic signs have been shown to be superior to conventional guide signs for some interchanges. The MUTCD encourages the highway agencies to continue with further experimentation using diagrammatic signs. The Wyoming State Highway Department conducted a study [3] to evaluate the diagrammatic signing system and to compare it with the current highway signing system. The main point of the investigation focused on diagrammatic sign comprehension and motorist response time behavior. The study used visual observations and motorist interviews to compare the situations before and after the change to the diagrammatic road signs. This study showed only slight improvements in the erratic movement metric. Zwahlen [4] analyzed video taped eye fixations and saccades (30 frames per second) for 32 young, healthy unfamiliar drivers, along rural two lane highways in Ohio under low beam illumination conditions at night for the approach to a curve/turn warning sign (curve/turn symbol) for two selected curves. The first-look distance (longitudinal distance measured from the sign to a driver's eyes at which a driver foveally fixates the sign for the first time), last-look distance (the distance measured from the sign to a driver's eyes where he/she moves the eyes away from the sign for the last time before reaching the sign), number of looks and durations of looks at the warning sign were of main interest in the study. The obtained eye scanning data were used to formulate two models that provide the Minimum Required Legibility Distance (MRLD). The MRLD for model 1 assumed that the last look time is independent of speed and model 2 assumed that the last look distance is independent of speed respectively. Cumulative last-look distance, first-look duration and last-look duration graphs were established and presented. The results of this study and a previous similar study indicate that drivers look on the average about two times at a warning sign during a night time low beam approach. It was found that between the first-look (information acquisition) and the last-look (confirmation) at a sign there was usually at least one eye fixation on the roadway ahead. Using cumulative eye fixation duration data obtained for straight road driving under low beam nighttime conditions published in another study and an average saccade duration of about.03 seconds, a sign reading distance model was developed which determines the distance (minimum required legibility distance, MRLD) at which a simple bold symbol on a warning sign must be recognized.

4 The model provides for a given speed the overall cumulative probability distribution function for the MRLD in terms of distance or in terms of time. The advantage of this model, which is applicable to warning signs with simple symbols under low beam illumination at night, is that it is totally based upon observed, recorded, and analyzed driver eye scanning and information-seeking behavior in the field. Warning signs, advisory speed signs, curve signs, stop ahead, and stop signs are some of the typical examples of ground-mounted signs. Zwahlen [5, 6, 7] reported that the driver eye scanning behavior is not much different for these signs that have different guidance or warning purposes. Zwahlen [5] analyzed warning signs that contained symbols and advisory speed signs and reported that drivers look at practically every warning sign and on the average look about 2.3 times at a warning sign. The average first look distance for daytime and nighttime were about 450 ft (137 m) from the warning signs. During nighttime both the warning sign sight distances were slightly larger and the fixation duration was considerably longer (nighttime average 0.75 seconds, daytime average 0.45 seconds). In another study Zwahlen [6] analyzed the advisory speed and curve signs and reported that for moderate curves, the drivers look about 2 times at a warning sign with fixations lasting seconds during daytime and at night drivers look 2.87 times with a fixation duration of around 0.6 seconds. The first look occurs at about 470 ft (143 m) during daytime and 530 ft (162 m) during nighttime. The average last look distance was around 260 ft (79 m) during daytime and 198 ft (60 m) at nighttime for moderate curves. When sharp curves are considered, the drivers look about 2 times at a warning sign with a fixation duration of around 0.5 seconds during daytime and the drivers look around 1.7 times with a fixation duration of 0.5 seconds during nighttime. The first look was at about 400 ft (122 m) during daytime and at about 260 ft (79 m) during nighttime. The average last look distances are about 220 ft (67 m) during daytime and 163 ft (50 m) during nighttime. Zwahlen [7], in a study on stop ahead and stop signs found that their effect on driver eye scanning and driving performance are different when compared to other warning signs. Drivers appear to use the stop signs as a fixed stimulus reference point to adjust their speeds and judge the distances to the stopping point. The drivers look 2.4 times to a stop ahead sign during daytime and 2.1 times during nighttime. On the other hand, the average number of looks for stop signs was higher. Drivers look on the average 6 times at the stop signs during daytime and 3.8 times during nighttime. The first look distance was around 900 ft (274 m) during daytime and 670 ft (204 m) during nighttime. DESCRIPTION OF THE INTERCHANGES STUDIED The test sites were selected based upon Ohio Department of Transportation (ODOT) traffic crash reports and traffic congestion reports. With the exception of the test site at Plain City Georgesville (SR 142) that was added later by an ODOT panel, all test interchanges were selected as typical examples of various interchange configurations. The following test interchanges were used: 1. SR 315 Southbound with I Brice Road Northbound with I Georgesville Road Westbound with I Roberts Road Eastbound with I Hilliard Rome Road (Southbound) Interchange with I Plain City Georgesville Road (SR 142) Southbound with I 70. 4

5 Figure 2 shows a map view of the third interchange (Georgesville Road Westbound with I270) as a typical example. Dimensions of the signs installed at this interchange are given in Figure 3. The installation of the signs in the field is shown in Figure 4. Two diagrammatic signs were used and were located ½ mile (805 m) and ¼ mile (402 m) prior to the last point of the gore, where a driver can still gain access to the freeway entrance ramp. EYE SCANNING METHOD Subjects, Apparatus, and Procedure Six male licensed drivers were recruited to participate in the study. All subjects, between the ages of 22-42, were unfamiliar with the Columbus, Ohio area (site of the experimental interchanges). One subject wore contacts during the eye scanning runs. This did not adversely affect the ASL 4000 eye tracking system performance. Subjects were advised that they were in command of the vehicle and were responsible for their actions while operating the vehicle. They were informed that the aim of the study was to determine how drivers dealt with the traffic and/or road situations in urban settings. They were not told the exact nature of the study, or about the experimental signs. Also, they were instructed to drive as normally as possible, obey all speed limits, and navigate the vehicle safely to the experimental destinations. At all sites the subjects were always started out in the wrong approach lane necessitating a lane change as they obtained the information from the diagrammatic signs. Driver eye scanning measurements were collected at night using an ASL4000 measurement system installed in a 1994 Ford Taurus station wagon. The subject drove the experimental vehicle shown in Figure 5a and wore a helmet apparatus designed to videotape and digitize what the driver was seeing as reflected from the transparent visor as shown in Figure 5b. Each subject made six eye scanning runs, one run each approaching the entrance ramps of each of the six experimental interchanges equipped with the diagrammatic signs. All runs were performed during nighttime due to a restriction of the ASL eye scanning system. Data Acquisition and Analysis Staging and calibration was always performed on a parking lot area near each corresponding interchange site. The eye scanning recording equipment was always new calibrated prior to performing a run towards an interchange site. The video data from the eye scanning runs were recorded on 8mm video cassettes. Recording of the eye scanning video and digital data always began well before the diagrammatic signs were visible and continued until the driver returned the vehicle back to the staging area. The analysis focused on the looking behavior of the subjects in the vicinity of the diagrammatic entrance ramp approach signs. A separate analysis was conducted for each diagrammatic sign location and for each subject. The first look and last look distance to each diagrammatic sign were carefully extracted from the video records. Also, the looking behavior was analyzed in order to determine the number and duration of looks at each diagrammatic sign. Data obtained in this study were compared to those from previous studies [4, 5, 6, 7] as a control. The initial step of the data analysis was to determine the relative location of each diagrammatic sign in question with respect to the distance counter shown on the video record. The exact longitudinal location of each diagrammatic sign thus became the origin of the longitudinal coordinate system for each video record. The distances at which looks to the 5

6 diagrammatic signs were recorded were always given as distances to the location of origin of the corresponding diagrammatic sign. Once the origin of a diagrammatic sign for a given video record (subject) was found, the video was stepped back and paused at the point, where the diagrammatic sign came into the field of view for the first time, usually a long distance ahead of the sign. The looks at the sign during the approach were then analyzed frame by frame. Each sign of interest had a virtual boundary illustrated as in Figure 6, extending the sign dimensions by approximately 15% in each direction. A fixation to a diagrammatic guide sign was defined as the duration from when the gaze crosshairs entered the virtual boundary to when the gaze crosshairs exited the virtual boundary. One or two frames where the crosshairs were outside of the virtual boundary did not terminate a fixation, provided that the crosshairs immediately returned to the inside of the boundary. In most cases, a fixation ended in an unambiguous saccade away from the sign area to another object (usually the road environment). This frame by frame analysis was necessary because the system does not automatically analyze the looks and the look duration on objects such as signs which do not remain fixed and of the same size in the video picture. In addition, an automatic analysis was not possible because the head movements were not tracked and, the eye tracker coordinate system was constantly moving with the subject s head. However, even if head movements were tracked, one could still not perform a completely automatic analysis, because the ASL proprietary algorithm that determines the start and end of a fixation does not consider moving objects such as a traffic sign seen from a moving car. The x,y gaze coordinates constantly change during an approach to a sign, even during a stable fixation on the sign. This coordinate change is due to smooth pursuit eye movements that are necessary to keep a selected feature on the sign fixated. With the technology present at the time this study was conducted, there seemed to be no better way than to analyze the eye scanning video record frame by frame. Each look was analyzed as follows. The numbers from the distance counter and the frame counter were recorded. The total duration of each look was calculated by subtracting the frame count at the end from that at the beginning of the look, and converting the number of frames into seconds (30 frames per second). Also, the distances at which each look began were calculated by subtracting the number on the distance counter when the look began from that at the location of the sign. These data were recorded separately for each interchange using a Microsoft Excel spreadsheet. RESULTS, DISCUSSION, AND CONCLUSIONS Because the diagrammatic entrance ramp approach signs were novel, much larger and different in appearance from the traditional trailblazer assemblies, particular interest was given to the number of looks that drivers made to the diagrammatic signs, the distance at which the first and last looks occurred, and the duration of the first and last looks. It is postulated that traffic signs with text or diagrammatic information that elicit more than an average of about four looks or eye fixations may be considered unacceptably distractive. The average number of looks for each diagrammatic sign is shown in Figure 7. An average of 2.8 looks were observed at the first diagrammatic sign, and an average of 2.0 looks were observed at the second diagrammatic sign (where present). This average number of looks indicates that the diagrammatic signs are noticed, processed and that they do not appear to unduly distract the drivers. The average number of looks found in this study are also very close 6

7 to the average number of looks and look durations for warning signs reported by Zwahlen [5, 6, 7]. The first and last look durations for the diagrammatic signs are shown in Figure 8a for the first sign and in Figure 8b for the second sign at each interchange. The observed sign reading times indicate a normal level of information processing and do not appear to be unduly long. Overall, the average first look durations were 0.74 seconds and 0.66 seconds for the first and second diagrammatic signs, respectively. The average last look durations were 0.68 seconds and 0.62 seconds for the first and second diagrammatic signs, respectively. These look durations fall well within the range of acceptability, and compare well with the sign reading times for warning signs given by Zwahlen in [4]. The first and last look distances are important dependent variables. With the help of the first look distance, one can determine approximately where a driver starts to extract visual information provided by a diagrammatic guide sign. The last look distance indicates approximately where a driver no longer pays attention to the sign. A diagrammatic sign must be designed and placed in such a way as to become completely legible between the first look distance and the last look distance. The box plots shown in Figure 9 and Figure 10 give a graphical representation of the first and last look distances for the experimental interchanges and the overall look behaviors. Figure 9 shows that the overall median first look to the first diagrammatic sign of the approaches occurs 123 m prior to reaching the diagrammatic sign, and that the last look occurred 48 m prior to the sign. The overall median first look distance to the second sign at the approaches shown in Figure 10 is at 104 m, which was somewhat shorter than the corresponding distance at the first sign. The second diagrammatic sign most likely served more as a confirmation of the first sign. (which is exactly the intent from a traffic engineering point of view). Thus, less processing was expected and quantitatively observed resulting in shorter average fixation durations and lower average number of looks for the second signs. In summary we may conclude that based on the eye scanning behavior results ground mounted diagrammatic signs on multi-lane arterials in advance of highway freeway interchanges do not unduly distract drivers and detrimentally affect a driver s looking behavior. Since such diagrammatic signs provide unfamiliar drivers with better advance navigational information (indicating the correct lane to enter the desired entrance ramp) it is suggested that such signs should be used on multi-lane arterials in addition to preexisting guide signs in cases where the cost of overhead span type sign bridges cannot be economically justified and additional advance guidance information to the motorists is highly desirable. In order to allow a driver more time for information processing and to execute a lane change maneuver, if required, such ground mounted diagrammatic signs should be located ½ mile (805 m) and ¼ mile (402 m) prior to the last point of the gore, where a driver can still gain access to the correct freeway entrance ramp. Further, the traffic volume and emissions would decrease since fewer drivers would miss their desired entrance ramps. More time for the drivers to select the proper lane will most likely result in fewer lane changing accidents, and smoother traffic flow. 7

8 8 REFERENCES 1. Zwahlen, Helmut T., and Schnell, Thomas, Evaluation of Ground Mounted Diagrammatic Entrance Ramp Approach Signs, Report for the Ohio Department of Transportation, Columbus, Ohio, October 2000, Report No. FHWA/OH-2000/018, pp US Department of Transportation, Manual on Uniform Traffic Control Devices for Streets and Highways, Federal Highway Administration, Wyoming State Highway Department, Evaluation of Diagrammatic Signing, Internal Report. 4. Zwahlen H.T., Traffic Sign Reading Distances And Sign Reading Times When Driving At Night, Transportation Research Record 1495, Transportation Research Board, National Research Council, Washington, DC., 1995, pp Zwahlen, H.T., Driver Eye Scanning of Warning Signs on Rural Highways, Proceedings of the Human Factors Society-25 th Annual Meeting, pp , Zwahlen, H. T., Advisory Speed Signs and Curve Signs and Their Effect on Driver Eye Scanning and Driving Performance, Transportation Research Record 1111, pp , Transportation Research Board, National Research Council, Washington, DC., Zwahlen, H. T., Stop Ahead and Stop Signs and Their Effect on Driver Eye Scanning and Driving Performance, Transportation Research Record 1168, Transportation Research Board, National Research Council, Washington, DC., 1988, pp

9 List of Figures 9 FIGURE 1. EXAMPLES OF ENTRANCE RAMP SIGNING FIGURE 2. MAP VIEW OF THE GEORGESVILLE ROAD (WESTBOUND) INTERCHANGE SITE WITH I FIGURE 3. DIMENSIONS OF GROUND MOUNTED DIAGRAMMATIC SIGNS INSTALLED AT GEORGESVILLE ROAD (WESTBOUND) INTERCHANGE WITH I 270. (1 IN = 2.54 CM; 1 MILE = 1609 M) FIGURE 4. GEORGESVILLE ROAD (WESTBOUND) INTERCHANGE WITH I FIGURE 5. EXPERIMENTAL VEHICLE AND EYE SCANNING EQUIPMENT USED IN THIS STUDY FIGURE 6. VIRTUAL BOUNDARY USED FOR ANALYSIS OF EYE FIXATION VIDEO RECORDS FIGURE 7. AVERAGE NUMBER OF EYE FIXATIONS PER DIAGRAMMATIC SIGN FIGURE 8. (A) AVERAGE DURATION OF EYE FIXATION (SECONDS) AT THE FIRST DIAGRAMMATIC SIGN OF EACH INTERCHANGE (B) AVERAGE DURATION OF EYE FIXATIONS (SECONDS) AT THE SECOND DIAGRAMMATIC SIGN (WHERE PRESENT) OF EACH INTERCHANGE FIGURE 9. BOX PLOTS OF LOOK (FIXATION) DISTANCES FOR FIRST DIAGRAMMATIC SIGN FIGURE 10. BOX PLOTS OF LOOK (FIXATION) DISTANCES FOR SECOND DIAGRAMMATIC SIGN... 19

10 10 a. Trailblazer assemblies. b. Diagrammatic sign. Figure 1. Examples of entrance ramp signing.

11 11 N Figure 2. Map view of the Georgesville Road (Westbound) interchange site with I 270.

12 12 a. Dimensions of Diagrammatic at ½ Mile b. Dimensions of Diagrammatic Sign at ¼ Mile Figure 3. Dimensions of ground mounted diagrammatic signs installed at Georgesville Road (Westbound) interchange with I 270. (1 in = 2.54 cm; 1 mile = 1609 m)

13 13 a. Diagrammatic Sign at ½ Mile (805 m) b. Diagrammatic Sign at ¼ Mile (402 m) c. Trailblazer Assembly d. Cantilever Overhead Signs Before and at First Gore e. Ground Mounted Sign at Second Gore Figure 4. Georgesville Road (Westbound) interchange with I 270.

14 14 Telescoping visor arm CCD pupil camera Optics module Subject sees through visor Infrared illumination and picture of pupil reflected on special coating Scene camera wiring harness a. Experimental Vehicle, 1994 Ford Taurus Wagon with Lane Tracking and Side View Cameras CCD scene camera (direct mode) b. ASL 4000 Eye Scanning Helmet Figure 5. Experimental vehicle and eye scanning equipment used in this study.

15 15 Virtual Boundary of Sign, Extending Sign Dimensions by 15% in each Direction Diagrammatic Sign Figure 6. Virtual boundary used for analysis of eye fixation video records

16 Diagrammatic Sign #1, Ave.=2.82 Diagrammatic Sign #2, Ave.= Number of Looks Brice, n=6 Georgesville, n=4 Roberts, n=6 US315, n=6 SR142, n=6 Hillard, n=6 Interchange (Note: not all interchanges contained two diagrammatic signs.) Figure 7. Average number of eye fixations per diagrammatic sign

17 Diagrammatic Sign #1, First Look, Ave.=0.74s 1 Diagrammatic Sign #1, Last Look, Ave.=0.68s Average Duration of Looks Brice First N=5 Last N=6 Georgesville First N=4 Last N=4 Roberts First N=6 Last N=6 US315 First N=5 Last N=5 SR142 First N=5 Last N=5 Hillard First N=6 Last N=6 Interchange (a) Average duration of eye fixation (seconds) at the first diagrammatic sign of each interchange Diagrammatic Sign #2, First Look, Ave.=0.66s Diagrammatic Sign #2, Last Look, Ave.=0.62s 1.05 Average Duration of Looks Brice Georgesville First N=4 Last N=4 Roberts US315 First N=5 Last N=5 SR142 First N=5 Last N=5 Hillard First N=4 Last N=4 Interchange (b) Average duration of eye fixations (seconds) at the second diagrammatic sign (where present) of each interchange. Figure 8. (a) Average duration of eye fixation (seconds) at the first diagrammatic sign of each interchange (b) Average duration of eye fixations (seconds) at the second diagrammatic sign (where present) of each interchange.

18 18 Diagrammatic Sign is Example Only Lower limit: Q1 1.5(Q3 Q1) Q1 Median Q3 Upper limit: Q (Q3 Q1) Box Plot Outlier Grouping Variable(s): Intersection Row exclusion: summary.svd LL LL D- D- 1: Overall, N=31 LL 1: z D- US LL 1: 31 US315, N=4 (1 No Looks, 1 Video Problem) D- SR 5 1: SR142, N=5 (1 Video Problem) LL 14 LL Ro D- 2 Last Look Distances, D- ber Roberts, N=6 1: Diagrammatic Sign #1 LL 1: ts Hill D- Ge Hillard, N=6 ard 1: org Bri es Georgesville, N=4 (4 Subjects Total) ce vill Ro e Brice, N=6 ad FL FL D- D- 1: Overall, N=31 FL 1: z D- US FL 1: 31 US315, N=4 (1 No Looks, 1 Video Problem) D- SR 5 1: FL 14 SR142, N=5 (1 Video Problem) FL Ro D- 2 D- ber Roberts, N=6 1: FL 1: ts First Look Distances, Hill D- Ge Hillard, N=6 ard Diagrammatic Sign #1 1: org Bri es Georgesville, N=4 (4 Subjects Total) ce vill Ro e Brice, N=6 ad Meters Units Figure 9. Box plots of look (fixation) distances for first diagrammatic sign.

19 Diagrammatic Example Only 19 Box Plot Grouping Variable(s): Intersection Row exclusion: summary.svd LL LL D- D- 2: z LL 2: D- US LL 2: 315 D- SR 2: 142 LL Ro LL D- bert D- 2: s 2: Hill LL Ge ard D- org 2: esv Bric ille e Ro ad FL FL D- D- 2: z FL 2: D- US FL 2: 315 D- SR 2: 142 FL Ro FL D- bert D- 2: s 2: Hill FL Ge ard D- org 2: esv Bric ille e Ro ad First Look Distances, Diagrammatic Sign #2 Lower limit: Q1 1.5(Q3 Q1) Q1 Median Q3 Upper limit: Q (Q3 Q1) Outlier Hillard, N=4 (2 No Looks) Georgesville, N=4 (4 Subjects Total) Hillard, N=4 (2 No Looks) Units Meters Overall, N=18 US315, N=5 (1 No Looks) SR142, N=5 (1 Video Problem) Last Look Distances, Diagrammatic Sign #2 US315, N=5 (1 No Looks) Overall, N=18 SR142, N=5 (1 Video Problem) Georgesville, N=4 (4 Subjects Total) Figure 10. Box plots of look (fixation) distances for second diagrammatic sign.