Pedestrians’ estimates of their own visibility: A simple and effective
Stacy A. Balka, Johnell O. Brooksb,⁎, Nathan Kleinc, Jason Grygiera
aDepartment of Psychology, Clemson University, Clemson, SC, USA
bClemson University International Center for Automotive Research, Clemson University, Greenville, SC, USA
cDepartment of Psychology, Lander University, Greenwood, SC, USA
a b s t r a c t a r t i c l ei n f o
Available online 18 February 2012
Introduction: Research has shown that both pedestrians and drivers drastically overestimate pedestrians'
nighttime visibility (NHSTSA, 2008a, 2008b; Owens & Sivak, 1996) and fail to appreciate the safety benefits
of proven conspicuity aids. One solution is educational intervention (Tyrrell, Patton, & Brooks, 2004); howev-
er, the on-road assessment of its effectiveness is expensive and time consuming. Method: Experiment One in-
troduces a computer-based alternative to the field-based approach, successfully replicating the previous
study's trends among 94 students who either receive or do not receive an educational lecture. Experiment
Two utilizes the simulation's portability to determine if professional roadway workers have a more accurate
understanding of pedestrian conspicuity than students. Results: Results among 88 workers show they do not
significantly appreciate the advantages of effective retroflective material configurations or vehicle headlamp
settings, for example, any better than non-lectured students in Experiment One. Impact: The study's results
demonstrate the need for education among all pedestrians and the benefits of efficient testing methods.
© 2012 National Safety Council and Elsevier Ltd. All rights reserved.
In the United States alone, nearly 5,000 pedestrians are killed in
traffic crashes each year, accounting for 12.5% of all vehicle-related
deaths (National Highway Traffic Safety Administration [NHTSA]
[NHTSA], 2008a, 2008b). This percentage is even higher elsewhere;
42% and 51% in Asia and the Middle East, respectively (Rumar,
2001). It is anticipated that by 2020 vehicle crash fatalities will be
the third leading cause of years of life lost (Peden et al., 2001). Despite
a reduction in both vehicle and pedestrian traffic, over half of all pe-
destrian fatalities in the United States occur at night (NHTSA). Poor
nighttime visibility has often been cited as a key causal factor of this
discrepancy (Owens & Sivak, 1996; Sullivan & Flannagan, 2002).
Several methodologies exist that have the ability to enhance visi-
bility and pedestrian conspicuity at night. Many of these approaches,
however, are expensive and difficult to implement (e.g., increased
street lighting, night vision enhancement systems). An alternative ap-
proach is to modify nighttime pedestrian apparel. It has been well
established that utilizing retroreflective material can dramatically in-
crease object visibility (e.g., NHTSA, 2001). Retroreflective material is
especially effective in enhancing nighttime conspicuity when placed
on the pedestrian's major joints: the ankles, knees, waist, wrists, el-
bows, and shoulders (Blomberg, Hale, & Preusser, 1986; Luoma,
Schumann, & Traube, 1996; Owens, Wood, & Owens, 2007; Sayer &
Mefford, 2004; Wood, Tyrrell, & Carberry, 2005). Marking these
major joints highlights both human form and motion (known togeth-
er as biological motion). This biological motion configuration of retro-
reflective material has been shown to generate pedestrian detection
distances significantly greater than pedestrians wearing no retrore-
flective material, those wearing a retroreflective rectangular vest,
and those wearing a traditional American National Standards Insti-
tute (ANSI) class II vest (e.g., Balk, Graving, Chanko, & Tyrrell, 2007;
Wood et al., 2005). These effects have been shown to be robust
even in the presence of visual clutter, visual distractions, changes in
pedestrian orientation, and both when pedestrians are walking and
standing still (e.g., Balk et al., 2007, 2008, Balk, Tyrrell, Brooks, &
Carpenter, 2008; Tyrrell et al., 2009).
Despite these impressive conspicuity advantages, the use of night-
time clothing incorporating biological motion has been rare among
pedestrians in both professional and non-professional environments.
One possible explanation for the underuse of this safety-enhancing
clothing is that pedestrians do not fully understand the magnitude
of the problem of their poor personal visibility. To nighttime drivers,
pedestrians not wearing conspicuity enhancing clothing are extraor-
dinarily difficult to see and recognize. At the same time, nighttime
drivers are acutely unaware of the extent to which their own vision
is degraded (Leibowitz & Owens, 1977). Thus, it is not surprising
that drivers are often startled when they encounter a pedestrian at
night. In fact, one study found nearly 25% of drivers involved in
vehicle-pedestrian crashes reported that they heard the sound of
Journal of Safety Research 43 (2012) 101–106
⁎ Corresponding author at: Department of Automotive Engineering, 4 Research Drive,
Greenville, SC, 29607, USA.
E-mail address: email@example.com (J.O. Brooks).
0022-4375/$ – see front matter © 2012 National Safety Council and Elsevier Ltd. All rights reserved.
Contents lists available at SciVerse ScienceDirect
Journal of Safety Research
journal homepage: www.elsevier.com/locate/jsr
the collision prior to seeing the pedestrian (see Leibowitz & Owens,
1986). Yet to nighttime pedestrians, the headlights of vehicles are
so salient that recognizing the vehicle from a great distance is a sim-
ple task. This may result in pedestrian overconfidence of drivers’ visu-
al abilities. Pedestrians may, then, not only fail to take actions to
enhance personal conspicuity, but also to assume that drivers will
be able to successfully take actions to avoid collisions with them.
This assumption could lead to pedestrians engaging in risky behav-
iors, such as, walking along dark shoulders or inappropriately cross-
ing the roadway in the presence of a moving vehicle.
It has clearly been shown that drivers have difficulty seeing and
recognizing pedestrians. Yet little work has focused on pedestrians’
beliefs of the distance at which drivers are able to see and recognize
them. Pedestrian beliefs play an important role in the actions they
may or may not take at night. Two early studies found that pedestrians
dramatically overestimate their own nighttime conspicuity (Allen et
al., 1970; Shinar, 1985). While this is an important finding, neither
study systematically manipulated pedestrian clothing, which modifies
pedestrian conspicuity. Tyrrell, Wood, and Carberry (2004) asked par-
ticipants to walk in place along the side of a roadway while wearing
either all black, all black with a reflective vest (a retroreflective rectan-
gle), all black with reflective material in a biological motion (“BioMo-
tion”) configuration, or all white. Participants were asked to estimate
the distance at which they judged they would be recognized by the
driver in a passing vehicle. Once again, pedestrians (participants) sig-
nificantly overestimated their actual conspicuity by a factor of 1.8x
(see Fig. 1). Pedestrians also estimated the shortest driver recognition
distances while wearing the all black clothing. However, driver recog-
nition estimates for the white, vest, and BioMotion configurations
were not significantly different. Furthermore, the BioMotion condition
was the only clothing in which pedestrians did not overestimate per-
sonal conspicuity; and in fact, pedestrians underestimated their con-
spicuity by a factor of 0.9x.
Based onthis evidenceitis clearthatpedestrians donotfully under-
stand the magnitude of the nighttime conspicuity problem. This lack of
understanding may lead pedestrians to act in an unsafe manner. One
possible way to address this problem is through education. Tyrrell,
Patton, and Brooks (2004) found that a single educational session can
be a very effective way to help pedestrians to better appreciate their
poor nighttime conspicuity. Two different groups of participants lis-
tened to two different lectures: the first heard a general topic lecture
about nighttime safety; the second heard a more specific and heavily
graphic lecture with several photographs and video clips. Experi-
menters then asked participants to walk toward a stationary vehicle
and indicate when they were “confident that the driver of the vehicle
can recognize you as a pedestrian.” Those hearing only the general lec-
ture produced distances 10%shorter thana controlgroup. Impressively,
those participants who heard a more specific and in-depth lecture esti-
mated visibility distances 56% shorter than the control group.
This study was the first to directly address the problem of nighttime
conspicuity and disproportionate pedestrian fatalities using an educa-
tional intervention. Tyrrell, Patton, and Brooks (2004) showed that ed-
ucation can successfully reduce the distance at which pedestrians
believe they can be recognized by drivers. This is an important step to-
ical next step is to modify and develop an effective educational
program. However, the methodologies used to test the effectiveness of
educational training on pedestrian estimates are often challenging; re-
quiring clear weather, dark hours, and the elimination of traffic. There-
fore,theprimary goalof Experiment1 in thepresentpaper isto testthe
ability to obtain results similar to those of Tyrrell, Patton and Brooks
(2004) using a simple computer-based application and to reduce the
distance at which participants believe that drivers will recognize them
as pedestrians. Furthermore, using this task, will participants generate
a similar pattern of estimated recognition distances as found in
Tyrrell, Wood, and Carberry (2004)? This technique is then used in Ex-
sional pedestrians, including –of particular interest– workers who have
extensive experience working on or near roadways at night.
2. Experiment 1
Data are reported from 94 university students (M=19.2 years). All
participants obtained at least 20/40 (6/12) vision and none reported
other visual pathologies. All participants had a valid driver's license
and a minimum of one year of driving experience (M=3.7 years). Par-
ticipants were recruited from several sections of an introductory psy-
chology course. One section of the course heard a lecture on nighttime
vision and driving(describedingreaterdetail below). A total of 40 peo-
ple who heard the lecture participated (Lecture group). The remaining
54 participants did not hear the lecture (Control group).
A subject matter expert gave a 75-minute night driving lecture to one
section of an introductory psychology class. None of the students in the
Lecture group were aware that the lecture was in any way related to an
was modeled after that presented in Tyrrell, Patton, and Brooks (2004).
The lecture emphasized visual physiology, the importance of visual con-
trast, the limitations of vision at night, and the possibility that we fail to
tive degradation” hypothesis was introduced as a way to understand the
neural physiology underlying nighttime vision as it applies to problems
associated with driving at night. In addition, two types of nighttime colli-
the broadside of a tractor trailer) and pedestrian-vehicle collisions. In-
cluded in the lecture were numerous photographs and video clips from
actual crashes or reconstructions of crashes. A video clip demonstrated
the importance of pedestrians’ clothing reflectance by showing four pe-
destrians each clad in different clothing configurations (black, white,
on the major joints) walking on the shoulder of a roadway at night. The
benefit of using retroreflective materials was also discussed. The lecture
included a summary of data describing pedestrian visibility distances as
measured by Wood et al. (2005).
At no point in the lecture was the present study mentioned, and
the lecturer was not present during data collection. Similarly, the
Fig. 1. From Tyrrell, Wood, and Carberry (2004). Pedestrians’ estimates of their own
conspicuity at night as a function of clothing condition and headlamp beam setting.
Each bar represents a mean (plus one standard error of the mean) of 10 younger and
10 older pedestrians.
S.A. Balk et al. / Journal of Safety Research 43 (2012) 101–106
experimenters who recruited the participants and collected the data
were not present during the lecture. This was done to eliminate any
apparent association between the lecture the students heard and
the study in which they were later asked to participate. A mean of
39 days (range=14 – 57 days) elapsed from when participants in
the Lecture group had heard the lecture to when data were collected.
Participants were shown four 8.5” x 11” pieces of paper, each with
a photograph of a male wearing one of four different clothing options
(smaller versions of these images can be seen in Fig. 2). Participants
were asked to make visibility estimates while imagining wearing
each of the clothing options.
1. Black. Black pants, black long-sleeve top, black gloves, and black
2. Tag (reflective rectangle). Black+a 30 cm x 17.5 cm (525 cm2)
chest mounted retroreflective rectangle. This is the same dimen-
sions as the Vest condition used by Tyrrell, Wood, and Carberry
3. BioMotion (biological motion). Black+retroreflective straps on
the wrists, elbows, shoulders, ankles, knees, and waist. The reflec-
tive material had a total surface area of 525 cm2(equivalent to that
of the Tag configuration).
4. White. White pants, white long-sleeve top, white gloves, and
2.1.4. Visibility estimate photographs
Participants were shown daytime photographs of a red 1998 Nis-
san Frontier pick-up truck at various distances. The location of the ve-
hicle ranged in distance from 10 to 1280 ft away from the viewer
(camera). Photographs were taken at 15 different distances (10,
14.1, 20, 28.3, 40, 56.6, 80, 113.1, 160, 226.3, 320, 452.5, 640, 905.1,
and 1280 ft). Fig. 3. demonstrates four of these images that partici-
pants viewed. The camera remained stationary in the center of the
lane of oncoming traffic (i.e., the vehicle of interest was in the center
of the photograph). A semi-rural two-lane flat roadway was used. No
extraneous vehicles were present. The photographs were displayed
digitally on a standard 15” LCD screen.
After obtaining informed consent, participants were given a brief
overviewof theexperimentand a demonstrationofhow retroreflective
material works using real material and a flashlight. Participants were
asked to imagine walking on a dark country road at night in the center
hicle traveling 45 mph (the posted speed limit) and were asked to
identify “the farthest distance that you think the driver of the vehicle
could recognize you as a pedestrian.” Participants selected this distance
by scrolling through the visibility estimate photographs on the laptop
screen. This was done eight separate times by each participant. Partici-
pants made estimates while imagining wearing each of the four cloth-
ing configurations two times: once while imagining the oncoming
vehicle was using its high beam headlamps and once while imagining
the oncoming vehicle was using its low beam headlamps. It should be
noted that daytime photographs were used due to the extreme difficul-
ty in attaining photographs in a nighttime environment that accurately
represent our nighttime perceptual experiences.
Throughout each trial, participants sat in a chair in front of the 15”
LCD screen on which the visibility estimate photographs were pre-
sented. A chin rest was utilized to maintain a gaze toward the center
of the screen. Each trial started on the photograph of the vehicle at
back and forth through the photographs using a standard tethered
mouse as many times as it was felt necessary. This scrolling methodol-
ogy created a “flip-book” effect, where when scrollingthrough thepho-
tographs, a video-like feel with attained. This technique also provided
an opportunity for participants to easily reverse and forward though
the images to decide which distance they thought was appropriate.
The photographs provided participants much simpler control over the
distances/images than what a movie clip might. That is, less time was
spent attempting to slow, speed, or reverse the static images than a
video might. Furthermore, this static image methodology eliminated
to tap into the distance of the vehicle alone without the influence of
time until potential impact. Participants were also asked to indicate if
they believed that the driver would be able to stop the vehicle in
enough time to avoid collision with the participant/pedestrian.
Participants also made a second set of visibility estimates while
imagining driving the vehicle. The results, however, were not signifi-
cantly different from those estimates made from the pedestrians’ per-
spective. As such, they are not included in the data reported here.
In both the Control group and the Lecture group, outliers were calcu-
of three data points from the Control group and one from the Lecture
group were removed from the data set (a standard deviation of >±3
was used as the criterion for exclusion). It should also be noted that
data from one participant from each group was not included in the data
set and over half of these participant's responses were outliers.
Fig. 2. Sample images of the clothing that participants were shown. From left to right, Black, Tag, BioMotion, White.
S.A. Balk et al. / Journal of Safety Research 43 (2012) 101–106
Overall, estimated recognition distances were greater when partici-
pants imagined the driver using high beams (242.8 ft; 74.0 m) than
when using low beams (158.2 ft; 48.2 m), F(1, 86)=110.09, pb.001.
The clothing pedestrians imagined wearing also significantly affected es-
timated recognition distances, F(3, 258)=90.44, pb.001. The BioMotion
configuration generated an estimated recognition distance (268.0 ft;
81.7 m) greater than Tag (213.3 ft; 65.0 m), White (202.0 ft; 61.6 m),
and Black (118.7 ft; 36.2 m). Black generated significantly shorter esti-
mated recognition distances than the three other clothing configura-
tions. Tag and White were not significantly different (see Fig. 4).
A significant interaction between imagined headlight beam and
clothing was found, F(3, 258)=5.50, p=.001. When looking at each
clothing configuration individually, the estimated recognition dis-
tances were significantly greater when imagining the oncoming driv-
er using high beams than when imagining low beam use (each
comparison, pb.001). When looking at high beam and low beams
separately, the pattern of responses distances among clothing config-
urations is the same as when the beam groups are combined (both
comparisons, pb.001; see Fig. 5).
Overall, a marginally significant difference between the Control
group and the Lecture group was found, F(1, 86)=3.3, p=.073. Specif-
ically, the Lecture group (182.8 ft; 55.7 m) gave shorter recognition
estimates than the control group (218.2 ft; 66.5 m). While this is only
marginally significant, when participants were asked if they thought
the oncoming vehicle could stop in time to avoid collision, there was
a significant difference between the groups, χ2(7)=14.93, pb.05. In
each clothing/beam combination fewer people in the Lecture group be-
lieved the vehicle would be able to stop than the Control group.
Additionally, a significant beam by participant group interaction
was found, F( 1,86)=4.31, p=.041. Both the Lecture group and the
Control group estimated significantly greater recognition distances
when imagining the oncoming vehicle utilizing high beams than
using low beams (pb.001). When examining the high beam data
alone, the Lecture group generated significantly shorter estimated
recognition distances (216.8 ft; 66.1 m) than the Control group
(274.0 ft; 83.5 m), F(1, 88)=5.31, p=.024. However, when looking
at low beams only, there was no significant difference between the
Lecture and Control groups, F (1, 88)=1.15, p>.05.
A significant interaction between clothing and participant group
was also found, F(3, 258)=10.98, pb.001. When looking at the Con-
trol group alone, clothing significantly affected estimated recognition
distances, F(3, 147)=36.84, pb.001. Estimates for Black (141.6 ft;
43.2 m) were significantly shorter than White (234.2 ft; 71.4 m),
Tag (243.1 ft; 74.1 m), and BioMotion (254.0 ft; 77.4 m). White, Tag,
Fig. 3. Four example vehicle distance photographs presented to participants.
Fig. 4. Participants’ (Experiment 1, Control group only) estimates of visibility distance
to an oncoming vehicle. Each bar represents a mean value (plus one standard error of
Fig. 5. Participants’ (Experiment 1, Lecture group only) estimates of visibility distance
to an oncoming vehicle. Each bar represents a mean value (plus one standard error of
S.A. Balk et al. / Journal of Safety Research 43 (2012) 101–106
and BioMotion were not significantly different from each other. When
looking at the Lecture group alone, clothing once again influenced es-
(282.0 ft; 86.0 m) were significantly greater than Tag (183.6 ft;
56.0 m), White (169.8 ft; 51.8 m), and Black (95.8 ft; 29.2 m). Black
generated significantly shorter distances than each of the other cloth-
ing configurations. Only White and Tag were not significantly differ-
ent (all comparisons pb.05).
Undergraduate students were asked to estimate the distance at
which they believed that a driver would be able to “recognize you
as a pedestrian” using a very basic computer based task using photo-
graphs. One group of participants heard a lecture on nighttime driv-
ing and nighttime driving crashes (Lecture group), while another
group of participants did not (Control group). Due to the nature of
the task it is important to focus on the pattern of results rather than
the physical distance estimates.
Much like the results of Tyrrell, Patton, and Brooks (2004) and
Tyrrell, Wood, and Carberry (2004), the Control group failed to appre-
ciate the conspicuity benefits that retroreflective material placed in a
biological motion configuration can provide. That is, the Control
group in neither of the studies understood that the BioMotion config-
uration would allow a pedestrian to be recognized as a pedestrian at a
greater distance than when wearing all white clothing. However, the
Control group in both studies did appreciate that all black clothing is
the least safe option and reported shorter recognition distances for
Black than for any other clothing conditions.
The current study provides further evidence that education does in-
deed help pedestrians to better understand the nighttime conspicuity
problem. While overall there was only a marginally significant differ-
ence between the Control group and the Lecture group, the Lecture
group better understood that different clothing configurations produce
different recognition distances. In other words, the Lecture group esti-
mated longer recognition distances for the BioMotion clothing than
the Tag, White, and Black clothing. This pattern is consistent with the
be able to stop the vehicle in time to avoid collision. Fewer members of
the Lecture group believed that the vehicle would be able to safely stop
in each of the clothing conditions than did members of the Control
group. Presumably, persons who believe that a vehicle would not be
able to avoid collision are more likely to take pro-active action to
move away from and avoid moving vehicles.
The current study demonstrated the value of educating people
about the nighttime pedestrian conspicuity problem. Additionally, it
provided an inexpensive alternative to quickly assess pedestrians’ be-
liefs regarding their own nighttime conspicuity. Nighttime data col-
lection requires not only an appropriate outdoor setting but is also
difficult due to participant recruitment, logistics, and weather condi-
tions. This simple computer-based task provides a much simpler and
more accessible manner to investigate pedestrians’ beliefs of their
personal nighttime conspicuity as well as the effectiveness of differ-
ent educational methods.
While Experiment 1 of the present study was successful, this – and
most other nighttime pedestrian conspicuity studies – focused on the
“common” pedestrian. This approach ignores an important group of
pedestrians – the professional pedestrian. It is possible that profes-
sional pedestrians (e.g., construction workers) may have a different
and possibly better understanding of nighttime conspicuity and the
benefits of different retroreflective clothing enhancements. Experi-
ment 2 focused on the distance at which roadway workers estimated
that they could be recognized by drivers when wearing clothing with
different configurations of retroreflective material.
5. Experiment 2
Data are reported from 88 transportation professionals, all of
whom had experience working on public roadways at night
(M=43.3 years; 19–62 years). Twenty percent of these participants
spoke Spanish as their native language. All participants reported nor-
mal or corrected-to-normal vision. All participants had a valid driver's
license and at least five years of driving experience (M=27.8 years).
The methodology/procedures in Experiment 2 are identical to that
of Experiment 1 with two exceptions. First, no participants heard the
lecture. Second, similarly to Experiment 1, participants were shown
8.5” x 11” pieces of paper, each with a photograph of a male wearing
a different clothing option. In addition to the Black, Tag, BioMotion,
and White clothing of Experiment 1, participants were also shown a
standard ANSI vest (similar to what most of the participants wear/
have worn in and along the roadway. Further, due to location con-
straints, a chin rest was not utilized. However, a string was used to
maintain gazing distance. All materials were translated to Spanish
and all data from Spanish-speaking participants were collected from
individuals who spoke the language fluently.
The same criteria for excluding outlying data were used in Exper-
iment 2 as in Experiment 1. A total of eight data points were removed
from the data set (a standard deviation of>± 3 was used as the cri-
terion for exclusion). It should also be noted that data from two par-
ticipants were not included in the data set – over half of these
participant's responses were outliers.
Similarly to the university students, roadway employees estimat-
ed greater recognition distances when imagining the driver using
high beams (308.98 ft; 94.18 m) than when using low beams
(222.16 ft.; 67.71 m), F(1, 82)=92.28, pb.001. The clothing partici-
pants imagined wearing also influenced estimated recognition dis-
tances, F(4,328)=55.36, pb.001. Much like both groups of the
university students, Black clothing generated the shortest estimated
recognition distances (143.50 ft; 43.74 m). Tag (256.51 ft; 78.18 m)
and White (277.58 ft; 84.61 m) also generated significantly shorter
estimated recognition distances than Vest (335.71 ft; 102.32 m) and
BioMotion (314.54 ft.; 95.87 m). Neither Tag and White nor Vest
and BioMotion were significantly different (see Fig. 6).
Further a significant interaction between vehicle beam setting and
clothing was found F(4, 328)=4.56, p=.001. This interaction can be
examined in different ways. If we look at the low beam data only,
there is once again a significant effect of clothing F(4, 336)=45.55,
pb.001. Similarly, if we look at high beam data only, there is a signif-
icant effect of clothing F(4, 336)=39.82, pb.001. In both cases, the
pattern of estimated response distance is the same as when beam
groups are combined. That is, Black generated the shortest distances,
Tag and White were shorter than Vest and BioMotion, and neither Tag
and White nor Vest and BioMotion were significantly different. We
can also examine this interaction by looking at each clothing configu-
ration individually. In every clothing condition, participants estimat-
ed significantly greater recognition distances when imagining
drivers using high beams than when using low beams.
Experiment 2 asked employees to estimate the distance at which
they believed that a driver would be able to “recognize you as a pe-
destrian” using the same computer-based simulation as Experiment 1.
S.A. Balk et al. / Journal of Safety Research 43 (2012) 101–106
It was found that workers did not fully appreciate the conspicuity
advantage differences provided by different clothing/configurations of
Previous work has not only shown that distributing retroreflective
material in a BioMotion configuration is effective in maximizing night-
time pedestrian conspicuity, but also that it is more effective than the
standard ANSI Class II vest at increasing drivers’ ability to see and recog-
nize pedestrians (Balk et al., 2007; Graving et al., 2009). However, it ap-
pears that most DOT workers are unaware of this information. The
employees, who often worked in hazardous nighttime road conditions,
ity advantages, yet research shows this assumption is incorrect. ANSI rec-
ommends that roadway workers wear these vests when working on or
near the roadway. Therefore, it would be logical for roadway workers to
assume that the recommended standard would provide maximal safety
(i.e., nighttime conspicuity). Furthermore, if persons working in or near
a roadway assume that they are maximally conspicuous to nighttime
drivers, they may engage in behaviors which place their safety at risk
8. General Discussion
Overall, asking participants to estimate the distance at which they
thought oncoming drivers could recognize them as a pedestrian using
a novel, simple computer-based task was successful. This method can
also be employed using standard printed photographs, if necessary or
so desired, making it even more versatile. This method produced a
similar pattern of results to those of using an outdoor nighttime envi-
ronment like those of Tyrrell, Wood, and Carberry (2004). While this
task is not intended to replace data collected outdoors, this computer-
based task has many advantages including time/monetary costs, con-
venience, and a manner in which educational interventions are able
to be rapidly tested. The current study also found that even those peo-
ple who have extensive nighttime exposure as a pedestrian are not
immune to misconceptions regarding the effects of different clothing
on conspicuity. This further emphasizes the importance of education.
Experiment 1 demonstrated that a simple educational session (75
minute lecture) was effective in helping participants to better under-
stand the nighttime pedestrian conspicuity problem as well as the ef-
fectiveness of different clothing configurations. Using the computer-
based task, future research will be able to quickly assess the efficacy
of different types/styles of nighttime pedestrian education. It will be
possible, then, to develop a viable, widespread educational tool
using these methods. This type of education can help people to better
understand the nighttime conspicuity problem and lead to a proac-
tive approach toward personal safety, which could in turn reduce
the overall rate of nighttime pedestrian fatalities. This is especially
relevant on a global scale, as there are an estimated 200,000 night-
time pedestrian fatalities each year worldwide (Rumar, 2001), and
automobile crashes are expected to become the 3 rd leading cause
of Years of Life Lost by 2020 (Peden et al., 2001).
Davis, Whitney Ford, Casey Jenkins, Katie Kimble, Tina Kubala, Kristin
ry Initiative and the Clemson University Research Grant Committee.
Allen, M., Hazlett, R., Tacker, H., & Graham, B. (1970). Actual pedestrian visibility and
the pedestrian’s estimate of his own visibility. American Journal of Optometry and
Archives of American Academy of Optometry, 47, 44–49.
Balk, S. A., Graving, J. S., Chanko, R., & Tyrrell, R. A. (2007). Effects of retroreflector place-
ment on the nighttime conspicuity of pedestrians. An open-road study. Presented at
the 2007 Annual Meeting of the Human Factors and Ergonomics Society, Baltimore, MD..
Balk, S. A., Tyrrell, R. A., Brooks, J. O., & Carpenter, T. L. (2008). Highlighting human form
and motion information enhances the conspicuity of pedestrians at night. Percep-
tion, 37, 1276–1284.
Blomberg, R., Hale, A., & Preusser, D. (1986). Experimental evaluation of alternative
conspicuity-enhancement techniques for pedestrians and bicyclists. Journal of
Safety Research, 17, 1–12.
Graving, J. S., Tyrrell, R. A., Balk, S. A., Mendel, J., Braly, N. M., Sinakhonerath, L., O'Hara,
L. H., & Moore, K. S. (2009). The effects of retroreflectivity and biological motion on
the visibility of pedestrians at night. Presented at the Presented at the 2009 Annual
Meeting of the Vision Sciences Society, Naples, FL.
Leibowitz, H. W., & Owens, D. A. (1977). Nighttime driving accidents and selective deg-
radation. Science, 197, 422–423.
Leibowitz, H. W., & Owens, D. A. (1986). We drive by night. Psychology Today, 20, 55–58.
Luoma, J., Schumann, J., & Traube, E. C. (1996). Effects of retroreflector positioning on
nighttime recognition of pedestrians. Accident Analysis and Prevention, 28, 377–383.
National Highway Traffic Safety Administration [NHTSA] (2001). The Effectiveness of Retrore-
flective Tape on Heavy Trailers. Washington, DC: U.S. Department of Transportation.
National Highway Traffic Safety Administration [NHTSA] (2008). Traffic Safety Facts
2007: Pedestrians. Washington, DC: U.S. Department of Transportation.
National Highway Traffic Safety Administration [NHTSA] (2008). Fatality Analysis Report-
ing System: General Estimates System, 2007 DATA SUMMARY. Washington, DC: U.S. De-
partment of Transportation http://www-nrd.nhtsa.dot.gov/Pubs/811003.PDF From,
Owens, D. A., & Sivak, M. (1996). Differentiation of visibility and alcohol as contributors
to twilight road fatalities. Human Factors, 38, 680–689.
Owens, D. A., Wood, J. M., & Owens, J. M. (2007). Effects of age and illumination on
night driving: A road test. Human Factors, 49, 1115–1131.
Peden, M. M., Krug, E., Mohan, D., Hyder, H., Norton, R., MacKay, M., et al. (2001). Five-
year WHO Strategy on Road Traffic Injury Prevention. Geneva: World Health Organi-
zation Ref : WHO/NMH/VIP/01.03.
Rumar, K. (2001). A Worldwide Perspective on Future Automobile Lighting Report NoUMTRI-
2001-35. Ann Arbor, MI: University of Michigan, Transportation Research Institute.
Sayer, J. R., & Mefford, M. L. (2004). High visibility safety apparel and nighttime conspi-
cuity of pedestrians in work zones. Journal of Safety Research, 35, 537–546.
Shinar, D. (1985). The effects of expectancy, clothing reflectance and detection criterion
on nighttime pedestrian visibility. Human Factors, 27, 327–333.
Sullivan, J., & Flannagan, M. (2002). The role of ambient light level in fatal crashes: in-
ferences from daylight saving time transitions. Accident Analysis and Prevention, 34,
Tyrrell, R. A., Patton, C. W., & Brooks, J. O. (2004). Educational interventions successful-
ly reduce pedestrians’ overestimates of their own nighttime visibility. Human Fac-
tors, 46, 170–182.
Tyrrell, R. A., Wood, J. W., & Carberry, T. (2004). On-road measures of pedestrians’ es-
timates of their own nighttime conspicuity. Journal of Safety Research, 35, 483–490.
Tyrrell, R. A., Wood, J. M., Chaparro, A., Carberry, T. P., Chu, B. -S., & Marszalek, R. P.
(2009). Seeing pedestrians at night: Visual clutter does not mask biological mo-
tion. Accident Analysis and Prevention, 41, 506–512.
Wood, J. M., Tyrrell, R. A., & Carberry, T. (2005). Limitations in drivers’ ability to recog-
nize pedestrians at night. Human Factors, 47, 644–653.
Stacy Balk is part of the SAIC Transportation Solutions Division located at Turner-Fair-
bank Highway Research Center.
Johnell Brooks Brooks is an assistant professor of Automotive Engineering at Clemson
University International Center for Automotive Research.
Nathan Klein is an assistant professor of Psychology at Lander University.
Jason Grygier was an undergraduate student at Clemson University.
Fig. 6. Participants’ (Experiment 2) estimates of visibility distance to an oncoming
vehicle. Each bar represents a mean value (plus one standard error of the mean).
S.A. Balk et al. / Journal of Safety Research 43 (2012) 101–106