Influence of Infrastructure Design on Distracted Driving (2022)

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.

17 Chapter 3. Summary of Known Relationships Between Infrastructure Elements and Distraction A literature review was conducted to identify studies that evaluated the relationship between distraction incidents and roadway infrastructure elements. The following summary provides background information on the causes of OVD in general but focuses on those related to infrastructure. 3.1 Outside-the-Vehicle Distractions A number of studies have explored the scope and scale of OVDs in general. These studies have included some distractions related to infrastructure. Instrumented vehicle studies have been used to evaluate the frequency or percentage of drivers who engage in OVD. In one study, vehicles were instrumented for 70 volunteer drivers in North Carolina and Pennsylvania, and driver behavior was coded for approximately 207 hours of video (Stutts et al. 2003). Around 86% of drivers engaged in some type of OVD, and OVDs made up about 1.6% of the driving time coded (an average of 3.2 external distractions per hour per driver). Although the source of the distraction could not always be determined from the outside- facing camera images, typical OVDs identified in the comment field included waving or talking to someone outside the vehicle, looking at houses or scenery, approaching toll booths, visiting drive-through windows at banks or fast food restaurants, observing work zone activity, simply looking out the side window at something, and reacting to bright sun glare. There were few recorded instances of drivers being distracted by pedestrians, children, or animals outside the vehicle. In comparison, an Australian NDS evaluated driver behavior using 70 drivers, with 3 hours of driving data collected per driver (McEvoy et al. 2006). A smaller percentage of drivers compared to Stutts et al. (2003), approximately 58%, was found to be engaged in an OVD, which included viewing outside people, objects, or events. The authors noted that men were more likely to engage in OVDs than women, and drivers aged 18 to 49 were more likely to engage in OVDs than drivers in older age groups. Data from the SHRP2 NDS were used to assess driver behavior for over 1,500 crash events (Dingus et al. 2016). Driver activities were coded from short video clips captured just prior to the crashes. Using a mixed effects random logistic regression model, Dingus et al. (2016) estimated that extended glances to external objects increased crash risk 7.1 times (95% CI = 4.8, 10.4). However, the study did not indicate how often the extended glances were the main factor in the crashes as opposed to indicating that the driver was paying close attention to something outside the vehicle that posed a crash risk. In structured interviews of 426 drivers, participants were asked to identify activities that drivers engage in that might distract them. Only 7% spontaneously thought of “looking at something outside of the vehicle.” However, 97% of drivers deemed looking outside the vehicle to be distracting, and, depending on age group and gender, 79% to 96% of drivers reported that they have been distracted in such a way (Prat et al. 2017).

18 A law enforcement survey was conducted in Virginia in which officers were asked to provide additional information about distractions when completing crash forms during a 4.5-month period (Glaze and Ellis 2003). About 2,800 surveys were collected with 2,900 distractions coded. The following relevant OVDs were noted: • Scenery or landmarks: 9.8%. • Weather conditions: 1.9%. • People: 1.0%. • Signs: 0.9%. • Lost or unfamiliar with road: 0.3%. • Objects in the roadway: 0.2%. • Other: 0.6%. Crash data and narratives have also been utilized to further investigate the role of OVDs on crashes. Gordon (2007) evaluated crashes in New Zealand (that occurred in 2002 and 2003) for which distractions were identified. In some crashes, more than one distraction was identified. It was found that 31% of fatal crashes, 43% of injury crashes, and 46% of all crashes involved an OVD. OVDs included the following: • 28% distracted due to sun strike. • 24% distracted by vehicles or focusing on vehicles when checking for traffic. • 14% looking at other vehicles. • 8% locating a destination. • 6% looking at people. • 5% looking at police, emergency vehicles, or other crashes. • 4% looking at landscape or architecture. • 4% looking at pedestrians or cyclists. • 3% dazzled by headlights. • 2% looking at animals. • 1% looking at other external events. • 1% looking at advertising signs. • 4% undefined. A similar breakdown of 221 OVD sources based on 744 crash narratives recorded in 1997 and 1998 in North Carolina was observed in a re-analysis by Wallace (2003a). Prevalent distraction sources included traffic or vehicles, people or animals in the roadway, police activities, and sunlight or sunset. The cause of the distraction was unspecified in more than a third of the cases. Data from the Australian National Crash In-Depth Study were used to identify distractions to evaluate the role of driver distraction and inattention in serious injury crashes (Beanland et al. 2013). The study included detailed surveys of crash participants. Data were available for 216 crashes. Around 2.8% of distractions were attributed to another road user’s behavior, 0.9% were attributed to street signs, and 0.9% were due to animals on the roadway. Looking specifically at fatal crashes, a study using crash reports from Norway found that OVDs were a possible contributing factor in about 1% of cases (Sundfør et al. 2019). In contrast, road

19 user inattention (e.g., failing to scan at an intersection) accounted for about 29% of cases. Inside the vehicle distractions were much more common; typical sources included mobile phones, vehicle systems such as GPS and backing cameras, passengers, and music or the radio. 3.2 Billboards and Outdoor Advertising Numerous studies have examined the effects of billboards on driver behavior. Not only are billboards the most-studied IRD; the studies on their effects span the greatest time range, from the 1950s to the present. In general, the presence of outdoor advertising/billboards has been associated with an increase in crashes or has had other negative impacts on driving. A quasi- induced exposure study by Sagberg (2016) that surveyed approximately 4,000 drivers involved in crashes in Sweden found the risk of being in a crash to be 6.5 times greater when drivers were distracted by advertising signs and boards. Oviedo-Trespalacios et al. (2019) conducted a systemic review of the literature relative to roadside advertising and driver behavior. In general, the authors noted a consistent finding that roadside advertising was correlated with a 25% to 29% increase in crash risk (as noted in their study based on (Sisiopiku 2015, Wallace 2003a, and Islam 2015)). A recent analysis of crash data from Israel examined the safety effects of billboards along the Ayalon Highway in suburban Tel Aviv, which were removed in 2008 and reinstalled in 2009 (Gitelman et al. 2019). The removal of the billboards was associated with a decrease in injury crashes of 30% to 40%, and their restoration was associated with an increase in injury crashes of 40% to 50%. Bendak and Al-Saleh (2010) conducted a driving simulator study with 12 drivers using two identical paths, one path that featured roadside advertising signs and one path that did not. The authors reported that instances of lane position problems and recklessness when crossing intersections were more frequent when roadside advertising signs were present. They also found an increase in instances of tailgating, speeding, and changing lanes without signaling when advertising signs were present. However, the differences were not statistically significant. Monash University Accident Research Centre conducted a simulator study with more than 100 Australian drivers and found that when billboards were present, drivers drove more slowly, took more time to change lanes, and made more errors while making lane changes. Older drivers made more lane change errors in the presence of static roadside advertising signs, while younger drivers engaged in longer glances at roadside advertising than drivers 35 or more years old (Edquist et al. 2011). Particular characteristics of advertisements such as the number of words, emotional impact, human representation, and design may impact lane keeping and increase gaze duration (Marciano and Setter 2017, Schieber et al. 2014). The presence of billboards may impact driving performance by diverting attention away from the roadway task. A study from Latvia examined glance duration and electroencephalograph activity for drivers along a busy freeway on the outskirts of Žilna (Hudák and Madleňák 2017). The 6.3 km (3.9 mile) segment included 191 nonelectronic billboards, an average of one every 216 ft (most of which were installed illegally). Average glance duration was 0.44 seconds. In a few cases, drivers gave a second (or third) glance, typically for a total glance duration of 0.55 seconds. On average, drivers looked at 43 of the 191 billboards (22.5%). Analysis of the brainwave data indicated that the billboards tended to produce instantaneous excitation (increased alertness).

20 Most studies have focused on static billboards. However, there are strong indications that billboards with dynamic content (Figure 3) have the potential to impose involuntary demands on driver attention, particularly if the signs use animation, flashing or scrolling text, full-motion video, jump-cut video, or other special effects (Dukic et al. 2013, Domke et al. 2011, Wachtel 2008). Source: Royalbroil, Creative Commons Attribution-Share Alike 2.5 Generic license. Figure 3. Billboard with scrolling text. Changeable or moving messages have been found to be more distracting than static messages (Beijer et al. 2004, Decker et al. 2015, Herrstedt et al. 2013, and Misokefalou and Eliou 2012a). In particular, gaze duration is longer and more frequent with changeable messages. Additionally, a simulator study by Milloy and Caird (2011) found increases in crashes and an increase in response time when drivers were responding to a lead vehicle in the presence of digital billboards compared to static billboards and baseline conditions. Moreover, a 2014 study of eight digital billboards on Interstate highways in Alabama found that over a five-year period (2008 through 2012), the crash rate per mile was 29% higher on billboard approaches than on control segments, which were located immediately downstream of each billboard but did not themselves feature billboards (Sisiopiku 2015). Research generally indicates that digital billboards are more effective at attracting and retaining driver attention than static billboards. These differences are believed to be caused by a combination of factors, including human sensitivity to rapid changes in illuminance (e.g., a sudden change in background color when one advertisement is replaced by another), visual halo effects at times of the day when the luminance of the billboard exceeds that of the ambient lighting, and, in some cases, the use of full-motion video. Digital technologies such as light emitting diodes (LEDs) and plasma displays can produce notably greater luminance contrast ratios than traditional signage technologies based on incandescent lamps or passive reflection of ambient light. Higher luminance, use of LED technologies, and changing luminance have all been associated with holding a driver’s attention for longer periods (Birdsall 2008, CTC & Associates 2012, Herrstedt et al. 2017, Roberts et al.

21 2013). Another notable recent technological development has been the use of high-intensity projectors to display art or advertising on building walls, sidewalks, and similar surfaces (Martin Rendl Associates 2013), but no studies evaluating the effects of these display technologies on distraction were found. Both the European Union and Canada have recognized the distraction caused by digital billboards and have sought to minimize the negative safety effects by developing guidelines related to digital advertising. The Transportation Association of Canada’s (2015) guidelines suggest that jurisdictions regulate the intrinsic characteristics of the signs (e.g., minimum frame duration, transition time between frames, brightness, and animation) and placement of the signs (e.g., maximum intensity, minimum spacing, and proximity to traffic control devices). A project from the Conference of European Directors of Roads titled Assessing Distraction of Vehicle Drivers in Europe from Roadside Technology-Based Signage (ADVERTS) developed 10 recommendations related to the use of roadside advertising (Weekley and Helman 2019). The recommendations fall into four categories: fundamentals (e.g., not hindering views or being confused with roadway signs), location (e.g., avoiding complex driving situations), content (e.g., banning animations, avoiding content that causes distraction, and requiring advertisements to be concise and legible), and physical design (e.g., maximizing duration, brightness, and overall size) (Weekley and Helman 2019). Neither set of guidelines currently provide thresholds for use (e.g., a minimum frame duration) because the research is not yet available to support specific values. 3.3 Smaller Advertising Signs A recent Italian NDS used eye trackers to compare driver responses to billboards with driver responses to smaller commercial signs (Costa et al. 2019). With the exception of the billboards, most of the signs were less than 6 m2 (64.6 ft2, about the size of two U.S. standard sheets of plywood), and some were as small as 0.2 m2 (2.1 ft2). By law, all were mounted at least 3 m (11.8 ft) from road shoulders, at least 250 m (820 ft) from intersections, and away from curves. A total of 15 drivers participated in the experiment (10 men and five women), each driving a total of 30 km (18.6 miles) along both directions of a 15 km stretch of a two-lane undivided secondary road, which had a traffic volume of approximately 200 vehicles/day. About half of the route was urban with a 50 km/h (31 mph) speed limit, and the remainder was rural with a 90 km/h (56 mph) limit. A total of 154 signs were present along the drive, which Costa et al. (2019) divided into six categories. Table 2 presents the percentages of signs that drivers fixated on, along with the mean and standard deviation of the fixation time.

22 Table 2. Fixation percentage and duration for various types of advertising. Sign Type Percent Fixated Mean Fixation Duration Billboards 31% 0.33 ± 0.50 sec On-premises business identification signs (vendor signs) 23% 0.26 ± 0.41 sec Business name + arrow (single sign) 23% 0.30 ± 0.36 sec Business names + arrows (multiple signs on the same post) 16% 0.28 ± 0.34 sec Fuel price signs with LED digits 27% 0.26 ± 0.10 sec Portable display boards (e.g., “Restaurant €10.00 Menu,” “Open – Phone Charging Station”) 12% 0.23 ± 0.41 sec Overall 24% 0.30 sec Source: Costa et al. 2019. The vertical and lateral positioning of the signs affected fixation rates and duration, as did sign size, character height and readability, and the salience of the messaging. Text over 100 mm (3.9 in.) tall tended to receive longer glances, while text-intensive signs with small characters generally had low fixation rates. Signs on the driving side of the roadway were fixated on slightly more often than signs on the opposite side of the roadway. The statistical distribution of fixation intervals had a long tail. For example, 24.8% of fixations on a single sign had a total duration of more than 0.5 seconds, 16.1% exceeded 0.75 seconds, 9.8% exceeded 1 second, and 1.5% exceeded 2 seconds. 3.4 Traffic Signs Two studies generally found traffic signing to be related to distraction. As noted in a previous section on general OVDs, a survey of law enforcement officers in Virginia asked officers to provide additional information about distractions when completing crash forms during a 4.5- month period (Glaze and Ellis 2003). Around 2,800 surveys were collected, with 2,900 distractions coded. The authors noted that almost 1% of coded distractions were recorded as “looking at signs,” which includes looking at road signs and traffic lights. An Australian study by Beanland et al. (2013) found that around 0.9% of distractions were attributed to street signs, which is similar to the findings of Glaze and Ellis (2003). These results are contradicted to a degree by a recent Italian study that used eye tracking equipment to examine the glance duration resulting from 75 official traffic signs located along an 8.3 km (5.1 mile) route (Costa et al. 2014). These signs included regulatory, warning, route priority, and informational signs consistent with European standards. A total of 22 drivers participated in the experiment (15 men and seven women). The average glance duration was only 0.15 seconds, generally too short for anything except a subconscious response. Overall, drivers fixed their eyes on only 25% of the signs, with regulatory signs tending to receive more fixations than signs in the other categories. 3.5. Urban Clutter In some cases, a distraction may be caused by a group of objects that individually may not be problematic. Horberry and Edquist (2009) define visual clutter as a situation in which the driver

23 has too many objects to attend to, which may include objects obstructing the driver’s view, conspicuous objects that are not necessary to the driving task, and too many driving-related objects. Urban clutter is included in this definition because many of the features and objects that characterize visual clutter are located within the right-of-way (Figure 4). Additionally, urban clutter may include sign clutter (too many traffic signs within a given area), which causes distraction because drivers must spend time looking for the information they need to complete the driving task. Source: Schuminweb, Creative Commons Attribution-Share Alike 2.5 Generic license. Figure 4. Visual clutter in Breezewood, Pennsylvania (2006). The primary concern with visual clutter is that it hinders a driver’s visual search for a particular object (Edquist et al. 2007), but it may also cause a distraction in and of itself. The concept of visual clutter is not easily defined, and what constitutes visual clutter may vary based on driver characteristics such as age, driving experience, or familiarity with the location (Dewar and Olson 2007). In a driving simulator study by Ho et al. (2001), younger and older drivers were required to search for traffic signs within roadway scenes. The researchers reported that for older drivers, response time increased and more fixations were made under cluttered conditions. Additionally, older drivers made more errors and had more and longer eye fixations than younger drivers. Edquist et al. (2007) conducted focus groups with 54 Australian drivers aged 18 to 60 to assess the impact of visual clutter on level of attention. Drivers were asked to define visual clutter, describe its most common forms in the roadway environment, and describe how visual distraction affects their driving. The authors found a strong connection between perceived levels of visual clutter and the level of attention that drivers felt would be necessary to drive safely in a particular scenario. Drivers noted that they would be more stressed and likely to be distracted when high levels of visual clutter were present.

24 When too much information is available, the driver may become confused and have an inadequate amount of time to process the available information (Wallace 2003b). When the stimuli in the roadway environment exceed the driver’s mental processing capabilities (for example, if the driver must scan numerous objects at one location to identify visual information relevant to the driving task), the driver will feel stressed and may begin to subconsciously adopt strategies for limiting cognitive workload, such as ignoring peripheral objects. 3.6 Wind Turbines To meet energy demand in a net-zero-emissions scenario, it has been predicted that electricity generation from wind will need to increase from 0.5% of global energy production in 2016 to 13% by 2070 (Shell International 2018). The construction of wind turbines along roadway corridors has been proposed to balance a shift toward low-emissions energy sources with the desire to limit the amount of privately owned land required for wind turbine projects. The potential for distraction to passing drivers has been cited among the numerous concerns raised by wind energy project opponents (Chapman and Crichton 2017). While there is a large amount of research on the effects of wind turbine noise (and noise reduction techniques), only three studies exploring the potential effects of wind turbines on driving behavior were found. These studies are summarized below. All of the existing studies note the possibility that drivers will be distracted by blade rotation. A second potential source of distraction is shadow flicker (i.e., the moving shadows of rotor blades) (National Research Council 2007). The latter is said to be a rare occurrence because it requires several conditions to exist at the same time: a wind tower near a roadway, clear weather, a low sun angle, and a wind direction that orients the turbine such that the roadway is shadowed. Commercial off-the-shelf software has been developed to predict the likelihood of shadow flicker based on site geometrics. No studies evaluating the effects of shadow flicker on motorist behavior were found. It may be difficult to generalize the results of wind turbine distraction studies. As of 2019, commercial vendors offer turbines with outputs ranging from less than 1 kilowatt to at least 12 megawatts (12,000 kilowatts). Consequently, the visual effects of different types of wind turbines are likely to vary considerably. Potentially relevant characteristics include distance from the roadway, tower height, rotor diameter, rotational speed, and perhaps blade shape. For commercial installations, these characteristics vary from site to site based on design considerations such as typical wind speed and ease of access to the commercial electrical grid. As of 2019, the wind turbines sold for onshore commercial electric generation have rotor diameters ranging from 82.5 m (270 ft) to 170 m (558 ft), corresponding to power outputs ranging from 1.6 to 5.8 megawatts; models designed for offshore applications have rotor diameters of up to 220 m (722 ft) and outputs of up to 12 megawatts (Vestas Wind Systems 2019, GE Power and Water Supply 2012, Siemens Gamesa 2019). Industry sources indicate that wind turbines designed for commercial generation rotate at about 10 RPM, depending on wind speed and other factors.

25 A study by researchers at the University of Calgary used a full-scale driving simulator to study driver perception-reaction time effects for a series of small (15 kilowatt) wind turbines placed near the driving lanes of an urban freeway (Milloy and Caird 2011). The simulated driving environment was a corridor patterned after a portion of the Ontario 401 freeway in Toronto, Canada, with three lanes in each direction and a 100 km/h (62 mph) speed limit. Each turbine was 5 m (16.4 ft) in diameter, 12 m (39 ft) from the road edge, and 15 m (49 ft) from the next turbine. This line of wind turbines extended 0.5 km (0.3 miles). The model included 33 turbines, 17 with blades parallel to the roadway and 16 with blades perpendicular to the roadway. Three blade speeds were modeled: stopped, 60 RPM, and 500 RPM. Milloy and Caird (2011) found no significant difference in perception-reaction time between a scenario in which the turbines were modeled and a baseline scenario that omitted the turbines. Compared to the baseline scenario, mean speeds when the turbines were modeled were 1.9 km/h (1.2 mph) slower. No differences in lateral lane position were observed. Most drivers looked at the turbines and slowed slightly as they passed them, but the overall differences in driving behavior were small. Debriefs with test subjects indicated that they noticed the turbines but did not perceive them as an important part of the experiment. The authors noted that some of the visual attention directed toward the turbines may have been a novelty effect and that the results cannot be generalized to the much larger wind turbines used for commercial power generation. Two studies were conducted to evaluate the effects of different setback placements for eight larger (90 m hub height and 110 m rotor diameter) wind turbines along part of the N15 (now A15) freeway in the Netherlands (Alferdinck et al. 2012, De Ceunynck et al. 2017). The site for both studies was located in an industrial area of Rotterdam, where the available space between the freeway and a shipping canal was too narrow to provide the tower setback recommended in national guidelines, as shown in Figure 5. Specifically, the proposed turbines were about 26 m (85 ft) away from the edge of the pavement, whereas the guidelines recommended a 55 m (180 ft) setback.

26 Source: T.W. van Urk, Shutterstock. Figure 5. Wind turbines along a canal in the Netherlands. In the first study, Alferdinck et al. (2012) used a simulator to evaluate the potential placement of the turbines long the freeway. The study report was prepared only in Dutch and is unpublished. According to a 2017 English-language summary by De Ceunynck et al. (2017), the study found statistically significant but small-magnitude increases in the standard deviation of driving speed and lateral position as well as an increase in the amount of time drivers would gaze at the turbines when the turbines were placed at the proposed 26 m (85 ft) from the edge of the pavement instead of the official 55 m (180 ft) setback. There were no indications that the project presented an unacceptable road safety risk, so a positive implementation recommendation was made, and the turbines were installed at the 26 m (85 ft) setback. De Ceunynck et al. (2017) then compared traffic speeds, speed deviation, and lateral vehicle positions before and after installation of the turbines. Average speeds decreased by 2.24 km/h (1.4 mph) after the turbines were installed. The standard deviation of the minute-by-minute average speeds (which is not directly comparable to the standard deviation of speeds for a group of individual vehicles) increased by 0.21 to 1.33 km/h (0.13 to 0.82 mph). Mean lateral position was found to shift to the left. This was consistent with previous research indicating that drivers tend to shy away from roadside objects. Contrary to the group’s expectations, however, the lateral shift was 136 mm (5.4 in.) when the blades were parallel to the roadway but only 78 mm (3.1 in.) when the blades were perpendicular to the roadway. The turbines did not appear to affect lane choice, and no serious traffic conflicts were observed either before or after installation.

27 De Ceunynck et al. (2017) concluded that while the changes in speed and lateral position were statistically significant, the magnitude of the changes was small and unlikely to affect traffic safety. Notably, the decrease in traffic speeds could be expected to improve safety, perhaps offsetting any safety disbenefits from the increase in the standard deviation of speeds. 3.7. Railroad Crossings One study (Tung 2014) was found that evaluated general distractions in the vicinity of railroad crossings (the crossing itself was not treated as a distraction source). Recorded video was used to analyze driver behavior at five railroad crossings near Lincoln, Nebraska. Based on through the windshield observations, the researchers endeavored to determine the gender of each of the 4,485 drivers and found that men were overrepresented in the data set. Drivers who appeared to be distracted were assigned to one of three categories: visual (looking outside the vehicle but not for the purpose of crossing safety), manual (eating, drinking, using a cell phone, smoking, reaching for an object, grooming, etc.), and cognitive (talking to passengers in the vehicle). Tung found that approximately one-third (33.6%) of the drivers appeared to be distracted; of the distractions observed, 13.9% were visual distractions, 25.8% were manual distractions, and 52.3% involved talking to passengers. Thus, overall, about 4.7% of the drivers were deemed to be distracted by visual stimuli outside their vehicles. More visual distractions were observed when a railroad crossing was located within 250 ft of an intersection or when the roadway was wet or icy. The statistical analysis also found fewer visual distractions when there was an unattended vehicle near the crossing. Although the distractions were not specifically tied to railroad crossing structures, the study provides some evidence that crossings have been considered in the context of OVDs. 3.8 Pavement Condition No studies that directly linked distressed pavement to driver distraction were found in searches of five major literature databases (Engineering Village, Google Scholar, ScienceDirect, Transport Research International Documentation [TRID], and Web of Science). Nevertheless, several studies have developed quantitative relationships between poor ride quality and various physiological effects and safety outcomes. Ride quality is often quantified using the International Roughness Index (IRI), with values less than about 95 in./mile generally considered “good” and values less than about 170 in./mile generally considered “acceptable.” Pavement ride quality (often informally called “roughness”) is an important factor in driver and passenger comfort. Pavements with very poor ride quality can also affect vehicle stability, resulting in higher crash rates than pavements with better ride quality. Additionally, pavement rutting can contribute to wet-weather crashes due to hydroplaning. For example, a study based on data from Arizona and North Carolina (Mamlouk et al. 2018) identified a sharp increase in crashes when the IRI exceeded 210 in./mile or the rut depth exceeded 0.4 in. The study’s authors warned that these values need to be used with caution due to measurement differences in various states. In addition, the analysis did not appear to consider possible correlations between geometric design standards and pavement maintenance standards (e.g., relaxed standards for both design and maintenance of low-volume roads). The authors suggested that severely distressed or rutted pavement can be distracting to drivers, but the analysis did not include distraction metrics.

28 Researchers have used a number of methods to quantify the relationships between vehicle occupant discomfort and roughness, such as physiological measurements of drivers’ heart rate or eye blink rate. This line of research indicates that pavements with poor ride quality result in an increased cognitive workload for the driver. For example, heart rate variability has been used as a measure of driving tension and fatigue. Using data from 24 drivers on a mountainous low- volume road in northwest China, a recent study computed a function to relate the road’s IRI data to heart rate variability data collected using an electrocardiograph and suggested that drivers’ cognitive workload becomes unacceptable when the IRI exceeds 4.2 m/km (266 in./mile) (Hu et al. 2017). 3.9 Geometric Features Several studies have developed a correlation between drivers’ cognitive workload and horizontal curvature. A study in Texas that consisted of both test track and on-road components found that a driver’s visual demand was significantly higher when driving on horizontal curves with sharp radii (Wooldridge et al. 2000). Several other studies also found relationships between drivers’ cognitive workload and radius of curvature (Fitzpatrick et al. 2000, Tsimhoni and Green 2001, Easa and He 2006). However, no studies were found that correlate distraction to roadway geometry. 3.10 General Infrastructure A handful of studies have explored distractions related to various infrastructure elements not already discussed. Most notably, a NDS was conducted on the Attica Tollway, a 51 km (32 mile) route that encircles Athens, Greece (Misokefalou et al. 2016). Glance durations from 29 drivers (age 26 through 55) were observed for 69 roadside elements such as advertisements, traffic signs, overpasses, noise barriers, and commuter rail stations located in the median (see Figure 2 above). A total of 1,454 glance durations were analyzed. Table 3 summarizes the durations for each type of element. Table 3. Average glance durations for roadway and roadside elements. Category Element Average Glance Duration (sec)* Advertising Advertisement 0.86 Temporary Banner 0.77 Gas Station Sign 0.90 Road Element or Structure Variable Message Sign 0.99 Noise Barrier 0.94 Information Sign 0.63 Toll Plaza 1.05 Bridge or Overpass 1.44 Other Buildings 0.66 Commuter Rail Station 1.23 Other 1.20 * Zero-duration values excluded. Source: Misokefalou et al. 2016.

29 A notable finding from Misokefalou et al. (2016) is that the average duration of glances toward bridges and overpasses, commuter rail stations, toll plazas, variable message signs, and noise barriers exceeded the duration of glances toward outdoor advertisements. Two of the commuter rail stations attracted especially long glances, averaging 4.2 seconds at one station and 4.5 seconds at the other. Additionally, a toll plaza that serves as the entrance to the main roadway attracted glances that averaged 4.2 seconds. Overall, Misokefalou et al. (2016) found that the glance durations were similar for women and men but considerably shorter in tunnels than on the open road and considerably longer in daytime than in lowlight conditions (at night or in tunnels). Glance durations were also longer when traffic was light and when the travel speed was under 80 km/h (50 mph). The number of traffic lanes strongly affected glance duration: glance duration was about half as long on one- lane sections as on four-lane sections, while the glance durations for two- and three-lane sections were equal and fell about midway between those of the one- and four-lane sections. There was no significant difference in glance time for elements placed in front of the driver as opposed to those on the side of the road, but elements that were within 2 m (6.6 ft) of the edge of the road received longer glances than those mounted farther away, and large elements received longer glances than small ones. 3.11 Discussion The extent to which infrastructure elements contribute to distraction-related crashes is largely unknown. This is likely due to multiple factors. First, the topic has not been the focus of much research, resulting in a dearth of literature. Second, identifying the relationship using crash data is difficult. Many infrastructure elements are common (e.g., signs), making it difficult to compare crashes along roadway segments with the infrastructure element of interest (test segments) to locations without the feature (control segments). In order to address this difficulty, some studies have attempted to isolate crashes labeled as involving an OVD. The main challenge with this approach is that there is not a consistent method for identifying and defining an OVD on crash forms. Most of the OVDs mentioned in police reports appear to be ephemeral: children or animals near the roadside, emergency vehicles responding to calls, sun glare, and so forth. Driver self-reporting of OVDs appears also to be unreliable, with problems ranging from drivers who are unaware that they are distracted to drivers who blame a crash on a nonexistent OVD in an attempt to conceal a mistake or misbehavior. Moreover, a recent technical note from NHTSA observes that law enforcement officers sometimes fail to distinguish OVDs from actions inherent to the driving task, such as looking at traffic signs (National Center for Statistics and Analysis 2019). Similarly, some of the more theoretical research in the driving psychology literature uses the word “distraction” to describe a broad range of stimuli received by drivers, whereas the focus of BTS-09 is on stimuli that divert drivers’ attention to nonessential tasks. Since identifying a correlation between crashes and distraction due to infrastructure elements is challenging, the majority of studies attempting to determine a link between roadway features and distraction have used simulator studies and utilized surrogate measures such as glance duration or glance location.

30 The vast majority of the published research on IRDs centers on the effects of billboards. Small outdoor advertisements are mentioned in two studies and appear to have a similar but less intense effect. The high level of interest in outdoor advertising research appears to reflect a fundamental and longstanding conflict between the public safety interests of road users and the interests of advertisers seeking to deliver messages to road users. Taken as a whole, this body of work suggests that the technical features, physical positioning, and creative content that contribute to the ability of an advertising display to attract a viewer’s attention also contribute to its distraction potential and consequent adverse safety outcomes. In practice, government agencies have responded to safety and aesthetic concerns with a wide range of policies, ranging from Vermont’s nearly complete ban on outdoor advertising (10 V.S.A. § 495) to New York City’s regulations requiring Times Square buildings to have a minimum of 1,000 ft2 of illuminated signage for each 50 linear feet of street frontage (NYC Planning 2011). Meanwhile, a recent study from Greece (Misokefalou et al. 2016) indicates that various physical features of the built environment have distraction effects that are similar to, or greater than, those of billboards. These features notably include commuter rail stations in a tollway median, toll plazas, and variable message signs. Additional research on the impacts of features such as these appears to be needed. Since these features tend to be most concentrated in urban driving environments and previous research has shown that drivers associate cluttered environments with challenging driving conditions, research on the combined effects of multiple infrastructure distractors also appears to be warranted. Distraction caused by other roadside objects has not been thoroughly investigated. For example, a few studies have explored the distraction potential of rotating wind turbine blades, but the generalizability of these results has been hampered by rapid changes in wind turbine design, which affects visual profiles and setbacks from roadways. Potential distractors such as scenic views, distinctive roadside architecture, public art, flags, graffiti, and non-advertising signs have also received little research attention. Although it is known that distracted, inattentive, or sleepy drivers have an increased risk of running off the road when a sharp curve follows a long tangent roadway segment, almost no research was found on the relationships between distraction and the characteristics of roadside features such as drainage structures, bridge abutments, or railroad crossings. Pavements with very poor ride quality have been shown to affect drivers’ heart rate and eye blink rate, but no research was found directly linking ride quality to distraction per se. The effect of repetition on the distraction potential of roadside objects also appears to be largely unexplored in the existing research literature. Presumably, road users would acclimate to familiar environments. Perhaps roadside objects that are distracting to first-time viewers begin to fade into the background for regular commuters. Transportation agencies can potentially address some roadside distractions through design changes or remedial measures such as walls, fences, visual screens, or landscaping. If budgetary and political constraints allow, agencies can also purchase and remove existing roadside distractors. New distractors can, at least in theory, be limited through planning and zoning regulations, which may require close interagency coordination. National organizations can assist state and local agencies by developing recommended technical standards for the size, placement,

31 and luminance of potential distractors; limitations on dynamic content; and so forth. However, it is likely unfeasible that all roadside distraction sources can be eliminated. Moreover, human factors research suggests that doing so might not be desirable.