ArticlePDF Available

The Difference in Night Visibility between Shared Bikes and Private Bikes during Night Cycling with Different Visibility Aids

Authors:

Abstract and Figures

In recent years, bike sharing has increasingly spread across the world. Compared with personal bikes, shared bikes are uniform and have bright surfaces to help the public to find them easily. At the same time, unfamiliarity is still a problem for some users of shared bikes. Therefore, these features should be understood to improve the night visibility of cyclists and improve traffic safety. Our study tested and compared differences in night visibility using five types of visibility aids. The results showed two cognitive differences between cyclists and drivers. First, cyclists believed that using flashing lights or static lights would provide better visibility than other visibility aids. However, using a static light and reflectors showed better results in our research. Secondly, compared to private bikes, cyclists showed more confidence in the nighttime visibility of shared bikes, especially with retroreflective strips. But the behavior of drivers in our study did not support such differences. A post-experiment survey was conducted to explore such cognitive differences, and showed that unfamiliarity with these strips was a possible reason for driver unawareness. This study will aid policy makers in incorporating suitable visibility aids within bike-sharing programs. Further, this study includes helpful advice for cyclists in terms of improving their night visibility.
Content may be subject to copyright.
sustainability
Article
The Dierence in Night Visibility between Shared
Bikes and Private Bikes during Night Cycling with
Dierent Visibility Aids
Chengcheng Wu and Dawei Chen *
School of transportation, Southeast University, Nanjing 210096, China; 230179239@seu.edu.cn
*Correspondence: dw_chen@seu.edu.cn
Received: 24 October 2019; Accepted: 27 November 2019; Published: 9 December 2019


Abstract:
In recent years, bike sharing has increasingly spread across the world. Compared with
personal bikes, shared bikes are uniform and have bright surfaces to help the public to find them
easily. At the same time, unfamiliarity is still a problem for some users of shared bikes. Therefore,
these features should be understood to improve the night visibility of cyclists and improve trac
safety. Our study tested and compared dierences in night visibility using five types of visibility aids.
The results showed two cognitive dierences between cyclists and drivers. First, cyclists believed
that using flashing lights or static lights would provide better visibility than other visibility aids.
However, using a static light and reflectors showed better results in our research. Secondly, compared
to private bikes, cyclists showed more confidence in the nighttime visibility of shared bikes, especially
with retroreflective strips. But the behavior of drivers in our study did not support such dierences.
A post-experiment survey was conducted to explore such cognitive dierences, and showed that
unfamiliarity with these strips was a possible reason for driver unawareness. This study will aid
policy makers in incorporating suitable visibility aids within bike-sharing programs. Further, this
study includes helpful advice for cyclists in terms of improving their night visibility.
Keywords:
bike share; cycling safety; night-time visibility; cognitive dierence between cyclists
and drivers
1. Introduction
In recent years, bike sharing has increasingly spread across the world, with approximately
800 bike-sharing programs available in 2015 [
1
]. According to 2017 data from the local government in
Nanjing (China), the annual growth rate of bicycle-vehicle collisions (BVCs) reached 1.5%. At the same
time, there is emerging research on improving cycling safety, mostly focused on helmet use [
2
] and
operational cycling speeds [
3
]. Compared with private bikes, shared bike users have lower operational
speeds and are more reluctant to wear helmets [
4
]. While the former is likely to improve cycling safety,
the latter reduces cycling safety. The safety influence of using shared bikes needs further research.
One of the explanations provided for vulnerable road user accidents is their poor visibility to
other drivers [
5
7
]. Shared bike features show a possible dierence in night visibility. Shared bikes
commonly use a uniform and bright surface, so that the public can find them easily [
8
,
9
]. In China,
the most popular shared bikes are Mobike and Hellobike. The former uses bright orange and the latter
is blue and white. In Canada, Bike Share Toronto is green and black. Further, Bixi in Britain is green
and gray. However, studies on the dierence in visibility between shared bikes and private bikes
are scarce.
On the other hand, using visibility aids to increase visibility is equally important, especially
when cycling in low-light conditions [
10
,
11
]. Visibility aids can increase the distance at which car
Sustainability 2019,11, 7035; doi:10.3390/su11247035 www.mdpi.com/journal/sustainability
Sustainability 2019,11, 7035 2 of 11
drivers detect cyclists at night [
12
,
13
]. Currently, front reflectors, static lights, and flashing lights are
widely used to improve cyclist visibility at night [
14
16
]. Further, some countries and regions have
started placing legal requirements on shared bikes to improve cycling visibility at night. For example,
in Mainland China, night-time reflectors need to be added but there is no requirement for helmets
and headlights. In the United States, Singapore, Australia [
17
], and some other locations, headlights
are provided with shared bikes as mandatory requirement. Other promising visibility aids include
retroreflective markers. Due to drivers’ visual sensitivity to human motion patterns, when placing
retroreflective markers on users’ major moving joints (such as ankles, knees, wrists, etc.), drivers can
recognize the presence of users more frequently and at much longer distances [
18
,
19
]. Much research
has evaluated visibility during night cycling using dierent visibility aids [
13
,
20
], but there is little
research on the dierence between shared bikes and private bikes when using dierent visibility aids.
Therefore, the main objective of this study was to evaluate the dierence between shared bikes
and private bikes when using visibility aids under low light conditions. Specifically, the aim of this
paper was to:
1.
Evaluate the dierence in visibility between shared bikes and private bikes with five types of
visibility aids.
2.
Explore the dierence in behavior when using visibility aids. Cyclists’ estimation of their own
visibility is among the key influencing variables.
This study can contribute to achieving a better understanding of the visibility of shared bikes.
Our results could help policy makers and cyclists to equip shared bikes with suitable visibility aids.
Further, we can provide more reasonable advice on night cycling for cyclists.
2. Methodology
The experimental methods employed in this research are introduced in this section, including
each step of this experiment and its related information.
2.1. Participants
In total, 50 participants completed the experiment, with a gender ratio of 1.27:1.00 (male: female).
Two subjects aged 25 and 40, respectively, with motor vehicle driving licenses issued by China’s
public security authority, were selected as the drivers of our test vehicle. Both of the subjects drive
regularly—every week in the past two years—and have had no accidents in the last three years.
Recruitment
Oine leaflets and online advertising were both used to recruit participants (as bike riders) for
our experiment. A total of 81 persons responded. However, due to logistics and test requirements,
53 persons initially participated, with 50 successfully completing the test.
Demographics
1.
Data on gender, age, bike accidents, etc., for each participant were collected using a questionnaire
before the test, and the statistical results are shown in Table 1. All participants came from Nanjing,
China. Accidents were defined as incidents in which the participants were injured or caused an
injury to anyone else while riding a bike during the past year.
Sustainability 2019,11, 7035 3 of 11
Table 1. Sociological statistical information on participants.
Male 28
Female 22
Age
Max. 48
Mean 33.4
Min. 24
Number of accidents involved
Max. 2
Mean 0.4
Min 0
Bike preference (for typical use) Private bike 34%
Shared bike 66%
2.2. Driving Scenario
The layout of the test driving scenario is shown in Figure 1, where the driving course was a
400 m-long (1312 ft) section of straight road. The starting point was the departure point of bike
riders and the terminal was the departure point of the test vehicle. The starting distance of these
points between bike riders and the test vehicle was 400 m (1312 ft). The road width is 8 m, which is
divided into two 4 m-wide (13.12 ft) mixed lanes for both motor vehicles and non-motorized vehicles.
The driving road section was closed during the driving tests, in order to avoid any interference caused
by other pedestrians or other motor vehicles.
Sustainability 2019, 10, x FOR PEER REVIEW 3 of 12
Mean
0.4
Min
0
Bike preference
(for typical use)
Private bike
34%
Shared bike
66%
2.2. Driving Scenario
The layout of the test driving scenario is shown in Figure 1, where the driving course was a 400
m-long (1312 ft) section of straight road. The starting point was the departure point of bike riders and
the terminal was the departure point of the test vehicle. The starting distance of these points between
bike riders and the test vehicle was 400 m (1312 ft). The road width is 8 m, which is divided into two
4 m-wide (13.12 ft) mixed lanes for both motor vehicles and non-motorized vehicles. The driving road
section was closed during the driving tests, in order to avoid any interference caused by other
pedestrians or other motor vehicles.
Figure 1. Test driving scenario.
Illumination of the Experimental Scene
All experiments were conducted at least one hour after sunset. A handheld light-measuring
device was used to test the illumination conditions of the road section. We ensured that the
illumination parameter for each test was not greater than 20 Lux at a height of 1.5 meters.
Weather in the Experimental Scene
All experiments were conducted in dry weather.
2.3. Test Vehicles and Bikes
A 20-inch HITO brand black bike was used as the private bike, with a net weight of 11 kg. This
bike was suitable for riders with a height of between 140 cm and 180 cm.
Sharing-bikes from OFO company were chosen as our test shared bikes. The classical OFO 2.0
bike type was used, with the most commonly used type being a 26-inch family bike size. The color
was golden yellow (CMYK color value C3, M26, Y96, K0).
The vehicle used in our experiment was a Volkswagen (POLO 1.4 L, 2007 version), with height
of 1,465 mm, width of 1,650 mm, length of 3,916 mm, and wheelbase of 2,460 mm. It was a five-door,
five-seat car, and the color was sky blue (CMYK color value C40, M0, Y0, K0).
2.4. Experiment Design
Figure 1. Test driving scenario.
Illumination of the Experimental Scene
All experiments were conducted at least one hour after sunset. A handheld light-measuring device
was used to test the illumination conditions of the road section. We ensured that the illumination
parameter for each test was not greater than 20 Lux at a height of 1.5 meters.
Weather in the Experimental Scene
All experiments were conducted in dry weather.
2.3. Test Vehicles and Bikes
A 20-inch HITO brand black bike was used as the private bike, with a net weight of 11 kg. This bike
was suitable for riders with a height of between 140 cm and 180 cm.
Sustainability 2019,11, 7035 4 of 11
Sharing-bikes from OFO company were chosen as our test shared bikes. The classical OFO 2.0
bike type was used, with the most commonly used type being a 26-inch family bike size. The color was
golden yellow (CMYK color value C3, M26, Y96, K0).
The vehicle used in our experiment was a Volkswagen (POLO 1.4 L, 2007 version), with height of
1,465 mm, width of 1,650 mm, length of 3,916 mm, and wheelbase of 2,460 mm. It was a five-door,
five-seat car, and the color was sky blue (CMYK color value C40, M0, Y0, K0).
2.4. Experiment Design
The experiments were designed to test the influence of visibility under low-light conditions using
dierent visibility aids. At the same time, the dierence between private bikes and shared bikes was
analyzed to determine eects on night visibility.
Types of Visibility aids
The influence of the following five types of visibility aids was tested:
Static light: an LED bike light was used with a maximum brightness of 250 LM and a direct
radiation distance at night of 80 m.
Flashing light: The flashing light used in this study has the same lighting capability as the static
light and a flashing frequency of 0.5 times a second.
Front reflectors: Yellow hard reflectors were mounted at the front of bikes and the front of
the pedals.
Retro-reflective strips on the major moveable joints: Red retro-reflective strips produced by China
ONLINELOVE Company were used. The strip width was 1 cm (0.394 inches). Prism reflective
films were attached to the strips. According to the test data provided by the manufacturer,
the maximum visible reflection distance was 300 m (984.25 inch).
No visibility factors.
Experiment Steps
The participants were asked to perform independent trials on the two kinds of bikes, which were
equipped with one of five types of visibility aids. Each participant was required to complete 10 trials
(five on shared bikes and five on private bikes), as listed in Table 2. For each participant, the order of
these ten trials was random.
Table 2. Trials and independent variables.
No. The Types of Bikes Visibility Aids
1
Shared bike
Static light
2 Flashing light
3 Front reflectors
4 Retroreflective strips
5 No aids
6
Private bike
Static light
7 Flashing light
8 Front reflectors
9 Retroreflective strips
10 No aids
The participants were asked to ride to the end of the road section at a normal speed (around
15 km/h) in each trial. One experimenter assisted the participants to select and change dierent bikes
and visibility aids and followed the participants to record two position coordinates of the participants:
1.
The position coordinate of the bike rider at which the bike rider thought that the motor vehicle
driver could see them was recorded as bike position 1 (BP1).
Sustainability 2019,11, 7035 5 of 11
2.
The position coordinate of the bike rider (participant) at which the motor vehicle driver said that
he/she could see the bike rider was recorded as bike position 2 (BP2).
The other experimenter recorded two position coordinates of the test vehicle:
1.
The position coordinate of the motor vehicle at which the bike rider thought that the motor
vehicle driver could see them was recorded as vehicle position 1 (VP1).
2.
The position coordinate of the motor vehicle at which the motor vehicle driver said that he/she
could see the bike rider was recorded as vehicle position 2 (VP2).
Speed measuring devices and cameras (used to record the speed and position of bikes) were
equipped on both the test vehicle and bikes for the test trials. The data recording process is shown in
Figure 2.
Sustainability 2019, 10, x FOR PEER REVIEW 5 of 12
2. The position coordinate of the motor vehicle at which the motor vehicle driver said that he/she
could see the bike rider was recorded as vehicle position 2 (VP2).
Speed measuring devices and cameras (used to record the speed and position of bikes) were
equipped on both the test vehicle and bikes for the test trials. The data recording process is shown in
Figure 2.
Figure 2. Data recording process.
Test Trial
One staff member introduced the test procedures without describing the test purpose to reduce
potential interference caused by participants. The participants were asked to repeat the instructions
to sure that they fully understand the experiment. A written description of the detailed procedures
was provided to every participant before the test. We also required each participant to complete a
random training trial to become familiar with the test procedure.
In addition, the following two criteria were used to evaluate the validity of a recorded result:
1. Speed requirement: with an average speed of 15 km/h, the speed variation of a bike rider
should not be greater than 5 km/h. The speed variation of the motor vehicle should not be greater
than 10 km/h, given the required driving speed of 40 km/h.
2. Track requirement: the cyclist is required to ride the bike in the right lane and the distance to
the central line of the road should not be less than 2 m (6.56 feet). The driver should also drive the
motor vehicle in the appropriate lane and not cross the central line.
If tests failed to meet one or all criteria, the tests were required to be performed again. If four
consecutive tests completed by a participant did not qualify, this participant was removed from the
study.
3. Results
Data analysis was performed using three dependent variables:
The estimated recognition distance (Der), or the distance between cyclists and drivers when
the cyclists reported that they thought they were visible to drivers.
The actual recognition distance (Dar), or the distance between cyclists and drivers when the
drivers reported that they could see the cyclists.
The distance difference (DD), which means the difference between Der and Dar. A value of
α = 0.05 was used in all analyses to determine the significance of main effects.
The multiple-factor repetitive measurement and the analysis of variance were conducted on the
estimated recognition distances, actual recognition distances, and distance differences as functions of
the kind of bikes (2 levels: level 0 = private bikes and level 1 = shared bikes), visibility aids (5 levels:
level 1 = static light; level 2 = flashing light; level 3 = reflector; level 4 = retro-reflective strips on the
major moveable joints; level 5 = no visibility aids). Table 3 shows the least squares means of Der, Dar,
and DD for private bikes and shared bikes across all visibility aids.
Figure 2. Data recording process.
Test Trial
One stamember introduced the test procedures without describing the test purpose to reduce
potential interference caused by participants. The participants were asked to repeat the instructions to
sure that they fully understand the experiment. A written description of the detailed procedures was
provided to every participant before the test. We also required each participant to complete a random
training trial to become familiar with the test procedure.
In addition, the following two criteria were used to evaluate the validity of a recorded result:
1.
Speed requirement: with an average speed of 15 km/h, the speed variation of a bike rider should
not be greater than 5 km/h. The speed variation of the motor vehicle should not be greater than
10 km/h, given the required driving speed of 40 km/h.
2.
Track requirement: the cyclist is required to ride the bike in the right lane and the distance to the
central line of the road should not be less than 2 m (6.56 feet). The driver should also drive the
motor vehicle in the appropriate lane and not cross the central line.
If tests failed to meet one or all criteria, the tests were required to be performed again. If four
consecutive tests completed by a participant did not qualify, this participant was removed from
the study.
3. Results
Data analysis was performed using three dependent variables:
The estimated recognition distance (D
er
), or the distance between cyclists and drivers when the
cyclists reported that they thought they were visible to drivers.
Sustainability 2019,11, 7035 6 of 11
The actual recognition distance (D
ar
), or the distance between cyclists and drivers when the drivers
reported that they could see the cyclists.
The distance dierence (DD), which means the dierence between D
er
and D
ar
. A value of
α
=0.05
was used in all analyses to determine the significance of main eects.
The multiple-factor repetitive measurement and the analysis of variance were conducted on the
estimated recognition distances, actual recognition distances, and distance dierences as functions of
the kind of bikes (2 levels: level 0 =private bikes and level 1 =shared bikes), visibility aids (5 levels:
level 1 =static light; level 2 =flashing light; level 3 =reflector; level 4 =retro-reflective strips on the
major moveable joints; level 5 =no visibility aids). Table 3shows the least squares means of D
er
,D
ar
,
and DD for private bikes and shared bikes across all visibility aids.
Table 3.
Least squares means of estimated recognition distance, actual recognition distance, and distance
dierence in private bikes and shared bikes.
Types of Bikes Der Dar DD
Shared bikes 78.9722(±3.0916) a28.3100(±0.8167) 50.6622(±2.7873)
Private bikes 74.5960(±3.0916) 26.8667(±0.8167) 47.7293(±2.7873)
aLeast squares means (standard error).
3.1. Estimated Recognition Distance (Der)
It can be seen from Table 4that, across all kinds of visibility aids, the estimated recognition distance
of shared bikes was significantly higher than that of private bikes, with p=0.0071 and F(1,49) =8.39.
Table 4.
Dierences of Least Squares Means in Pairwise Comparison of estimated recognition distance
(shared bikes versus private bikes).
Group 1 Group 2 Estimate Dierences Standard Error DF t Value Pr > |t|
shared bikes private bikes 4.3762 1.5104 49 2.9 0.0071
For comparison of private bikes and shared bikes, as shown in Table 5, both static and flashing
lights provide significantly higher D
er
values than the use of reflectors, retro-reflective strips, and no
aids. No aids were the least visible scenario based on the shortest distance of detection for all tested
visibility aids. There was no significant dierence in D
er
between static light and flashing light, or
between reflector and retro-reflective strips. Table 5shows the dierences of Least Squares Means in
pairwise comparison.
Table 5.
The Dierences of Least Squares Means in Pairwise Comparison of estimated recognition
distance (visibility aids).
Visibility Aids 1 Visibility Aids 2 Estimate
Dierences
Standard
Error DF t Value Pr > |t|
static light reflector 14.6 2.3882 196 6.11 <0.001
static light retroreflective strips 15.2658 2.3882 196 6.39 <0.001
static light flashing light 3.6942 2.3882 196 1.55 0.1246
static light no aids 24.5612 2.3882 196 10.28 <0.001
reflector retroreflective strips 0.6658 2.3882 196 0.28 0.7809
reflector flashing light 10.9058 2.3882 196 4.57 <0.001
reflector no aids 9.9612 2.3882 196 4.17 <0.001
retroreflective strips flashing light 11.5717 2.3882 196 4.85 <0.001
retroreflective strips no aids 9.2954 2.3882 196 3.89 <0.001
flashing light no aids 20.8671 2.3882 196 8.74 <0.001
Sustainability 2019,11, 7035 7 of 11
The D
er
of shared bikes with a static light was also significantly higher than that of private
bikes, with p=0.0111. Dierently, there was no statistical dierence between flashing light, reflector
retro-reflective strips, or no aids (Table 6).
Table 6.
Dierences of Least Squares Means in Pairwise Comparison of estimated recognition distance
(shared bikes with visibility aids versus personal bikes with visibility aids).
Shared Bikes’
Visibility Aids
Personal Bikes’
Visibility Aids
Estimate
Dierences
Standard
Error DF t Value Pr > |t|
Light lamp light lamp 8.7167 3.3774 196 2.58 0.0111
reflector reflector 0.4833 3.3774 196 0.14 0.8865
retroreflective strips retroreflective strips 3.0883 3.3774 196 0.91 0.3624
flashing light flashing light 4.165 3.3774 196 1.23 0.22
no aids no aids 5.4275 3.3774 196 1.61 0.1108
3.2. Actual Recognition Distance (Dar)
In the tests of fixed eects, as indicated in Table 7, there was no statistical dierence between
private bikes and shared bikes for actual recognition distance, with F(1,49) =3.18 and p=0.0849 >0.05.
Table 7.
Dierences of Least Squares Means in Pairwise Comparison of actual recognition distance
(shared bikes versus private bikes).
Group 1 Group 2 Estimate Dierences Standard Error DF t Value Pr > |t|
Shared bikes Private bikes 1.4433 0.809 49 1.78 0.0849
As shown in Table 8, use of a bike with a reflector resulted in a higher actual distance than use
of a bike with a flashing light, retro-reflective strips, or no aids, with positive estimate dierences
(6.1708 for retro-reflective strips, 7.6375 for flashing light, and 7.35 for no aids) and values of p<0.0001.
Interestingly, use of both a static light and a reflector had a larger positive estimate dierence compared
to use of a flashing light (6.0292 for static light with p=0.005; 7.6375 for reflector with p<0.0001),
indicating that use of a flashing light reduced actual recognition distance compared to use of a static
light and reflector.
Table 8.
The Dierences of Least Squares Means in Pairwise Comparison of actual recognition distance
(visibility aids).
Visibility Aids 1 Visibility Aids 2 Estimate
Dierences
Standard
Error DF t Value Pr > |t|
static light reflector 1.6083 1.2792 196 1.26 0.2112
static light retroreflective strips 4.5625 1.2792 196 3.57 <0.001
static light flashing light 6.0292 1.2792 196 4.71 <0.001
static light no aids 5.7417 1.2792 196 4.49 <0.001
reflector retroreflective strips 6.1708 1.2792 196 4.82 <0.001
reflector flashing light 7.6375 1.2792 196 5.97 <0.001
reflector no aids 7.35 1.2792 196 5.75 <0.001
retroreflective strips flashing light 1.4667 1.2792 196 1.15 0.2539
retroreflective strips no aids 1.1792 1.2792 196 0.92 0.3585
flashing light no aids 0.2875 1.2792 196 0.22 0.8226
For use of the same retro-reflective strips, the actual distance was statistically higher with shared
bikes than that for private bikes, with a positive estimate of dierences of least squares means of 4.2417
(
±
1.809 standard error), p=0.0207. This dierence between shared and private bikes remained when
there was no visibility aid, with a positive estimate of 4.25 (
±
1.809 standard error) and p=0.0205.
When equipped with a flashing light, private bikes were visible at a greater distance than shared bikes,
Sustainability 2019,11, 7035 8 of 11
with an estimate of 4.1583 (
±
1.809 standard error) and p=0.233. Table 5shows the dierences of Least
Squares Means in pairwise comparison. (Shown in Table 9)
Table 9.
The Dierences of Least Squares Means in Pairwise Comparison of actual recognition distance
(Shared bikes’ visibility aids versus Personal bikes’ visibility aids).
Shared Bikes’
Visibility Aids
Personal Bikes’
Visibility Aids
Estimate
Dierences
Standard
Error DF t Value Pr > |t|
static light static light 0.2667 1.809 196 0.15 0.8831
Reflector reflector 3.15 1.809 196 1.74 0.0843
retroreflective strips retroreflective strips 4.2417 1.809 196 2.34 0.0207
flashing light flashing light 4.1583 1.809 196 2.3 0.0233
no aids no aids 4.25 1.809 196 2.35 0.0205
3.3. The Distance Dierence (DD) between Estimated Recognition Distance and Actual Recognition Distance
In the tests of fixed eects of DD, as shown in Table 10, there was no significant dierence for
dierent types of bikes (F(1,49) =3.31 and p=0.0793 >0.05). Given those results, we only compared
the DD under dierent visibility aids, as shown in Table 11.
Table 10.
The Dierences of Least Squares Means in Pairwise Comparison of distance dierences
(Shared bikes versus Private bikes).
Group 1 Group 2 Estimate Dierences Standard Error DF t Value Pr > |t|
Shared bikes Private bikes 2.9328 1.6125 29 1.82 0.0793
Table 11.
The Dierences of Least Squares Means in Pairwise Comparison of distance dierences
(visibility aids).
Visibility Aids 1 Visibility Aids 2 Estimate
Dierences
Standard
Error DF t Value Pr > |t|
static light reflector 16.2083 2.5496 116 6.36 <0.001
static light retroreflective strips 10.7033 2.5496 116 4.2 <0.001
static light flashing light 2.335 2.5496 116 0.92 0.3617
static light no aids 18.8196 2.5496 116 7.38 <0.001
reflector retroreflective strips 5.505 2.5496 116 2.16 0.0329
reflector flashing light 18.5433 2.5496 116 7.27 <0.001
reflector no aids 2.6112 2.5496 116 1.02 0.3079
retroreflective strips flashing light 13.0383 2.5496 116 5.11 <0.001
retroreflective strips no aids 8.1162 2.5496 116 3.18 0.0019
flashing light no aids 21.1546 2.5496 116 8.3 <0.001
As shown in Table 11, the use of static light gave a positive estimate of distance dierences
compared to the use of a reflector, retro-reflective strips, or no aids (16.2083 compared to a reflector,
with p<0.001; 10.7033 compared to retro-reflective strips, with p<0.001; and 18.8196 compared to no
aids, with p<0.001). These results indicated that cyclists overestimate their visibility to a greater degree
when using a static light than when using a reflector, retro-reflective strips, or no aids. At the same
time, the results for night cycling without any visibility aids indicated less overestimation of cyclist
visibility, with a lower DD value using no aid compared to that using a flashing light, retro-reflective
strips, or static light.
Additionally, for both types of bikes and all kinds of visibility aids, the results showed a positive
estimate value for DD (49.1958, with a standard error of 2.6682, p<0.001) in the solution for fixed eects,
which indicated a statistically higher estimated recognition distance relative to the actual recognition
distance. In other words, night-cyclists might overestimate their visibility in low light conditions, as
reported by Wood [
18
,
19
]. We also did a one-sample T-test for the hypothesis H0 that the ensemble
Sustainability 2019,11, 7035 9 of 11
average of DD is zero, and the result showed that t (49) =39.8296 and p<0.001, which allowed the
rejection of H0.
3.4. A Post-Experiment Survey
Analysis of the experiment data in Section 3.3 revealed that although night-cyclists have greater
confidence when using a bike with a flashing light, our experiment indicated that static light and
retro-reflective strips have better night visibility. Because few bikes in China are equipped with lights,
especially flashing lights, it is necessary to assess the familiarity of drivers with flashing-light bikes.
We next did a survey to probe cyclists’ views about private bikes and shared bikes in a night-cycling
scene with dierent visibility aids. We then used a thirty-second video to assess drivers’ familiarity
with flashing-light bikes. The survey was completed by 187 respondents who ranged in age from 20 to
59. Of the respondents, 123 people were invited to answer questions as cyclists, and 65 people were
invited to complete this survey as drivers.
In the survey of cyclists, 27.6% interviewees said that they felt a perceptible dierence between
using a shared bike or a personal bike. The rest (72.4%) reported no or only a slight dierence between
using a shared bike or a personal bike. Almost all interviewees considered shared bikes to be more
visible for night cycling (93.5%). In response to a question on which type of visibility aid is the most
visible for night cycling, 91.9% interviewees selected a flashing light, and the remaining respondents
chose a static light.
The survey of drivers did not include written questions. Drivers watched a thirty-second video
that was recorded from the driver’s perspective about a bicyclist cycling at night from the opposite
direction (nearly 200 meters away) on a bike with a flashing light. As they watched this video, we
asked what they thought the flashing light was. No drivers thought it was a bike during the first 10 s
of the video and 41.5% drivers reported that it might be a bike in the middle 10 s. After watching the
whole video, 81.5% of viewers thought it was a bike, but only 61.5% were sure. This result indicated
a low ability of drivers to recognize a bike with a flashing light, which may explain the dierence
between our result and that of Wood [18].
4. Discussion
We have evaluated the visibility dierence between private bikes and shared bikes with five
types of visibility aids. The results showed cyclists overestimate shared bikes’ visibility in lowlight
conditions. As a visibility aid, a flashing light does not provide as much visibility to others as cyclists
expect it to. The diculty of quick recognition one of the main the possible causes of this, as seen in
the post-experiment survey.
Firstly, one result in this research is consistent with Wood [
18
,
19
], which indicated that cyclists
overestimate their own night-time visibility. One possible explanation for this is that cyclists tend to stand
on their own shoes to estimate recognition distance. A previous study showed that cyclist-motorists
had fewer collisions with cyclists and detected them at a greater distance [
5
]. According to bike crashes
research in New Zealand conducted by Tin et al. [
21
], cyclists’ visibility may be improved if more
people cycle and fewer people drive cars. In other words, cyclists or cyclist-motorists may have a
longer recognition distance for other cyclists than motorists. So in our research, if our cyclist tends to
estimate their recognition distance as a cyclist observing another cyclist, it would be a shorter estimate
than a motorist would actually take to spot a cyclist.
One interesting result is that cyclists think shared bikes are more visible than private bikes while
there is actually no significant dierence. Dierent proportions of sensory visibility and cognitive
visibility between cyclists and motorists are one possible reason for this. Sensory visibility shows the
extent to which an object can be distinguished from its environment, due to its physical characteristics
such as its shape, brightness, color, and angular size [
22
,
23
]. Obviously, shared bikes have a higher
sensory visibility than private visibility due to its bright and uniform appearance. Cognitive visibility
refers to the extent to which an object be noticed due to an observer’s experience, expectations,
Sustainability 2019,11, 7035 10 of 11
and objectives [
24
27
]. In lowlight conditions, it can assume that shared bikes and private bikes may
not have a significant cognitive dierence. This means motorists may obtain the same observation
times, no matter whether they encounter a sharing bike or private bike moving in the opposite
direction. And they will extract similar information from trac scenes based on their own expectations.
Cyclists may estimate their own bikes’ visibility mostly based on sensory visibility, while drivers
mostly use cognitive visibility. However, for now this remains a hypothesis, which targeted research
should explore further in the future.
Another cognitive dierence is that flashing lights do not provide as good a visibility as cyclists
estimate. A static light and flashing light showed a higher estimated recognition distance than reflector
and retroreflective strips. But a static light and reflector had a significantly higher actual distance
than flashing light, retroreflective strips, and no aids. A flashing light still provides higher sensory
visibility, due to our visual sensitivity towards patterns of human motion [
28
,
29
]. However, it may
be hard for drivers to recognize, which means it provides bad cognitive visibility, according to the
post-experiment survey.
However, there is still a gender gap in using shared bikes. This could be a constraint for the
validity of similar experiments if these gaps are not previously addressed. The gender distribution of
cyclists and the gender distribution of drivers were not representative of the gender distribution of
these two groups in the surveyed area. The experimental sample in our research included more males
than females. It would be interesting to carry out the same experiment with a dierent sex ratio to
study the eect of gender on cyclists’ and drivers’ estimation dierences between shared bikes and
private bikes.
We evaluated the visibility dierence in shared bikes and private bikes using five types of visibility
aids. The current study can advise policy makers in providing proper visibility aids and improving the
safety of share bike programs. In addition, it suggested that there is a dierence in sensory visibility and
cognitive visibility between cyclists and motorists. Further research on cyclists and drivers’ sensory
visibility and cognitive visibility is necessary for improving cyclists’ visibility in lowlight conditions.
Author Contributions:
C.W. and D.C. conceived and designed this research. C.W. and other five graduate
students completed these experiments and collected the data. C.W. wrote the paper. D.C. reviewed and edited the
manuscript. All authors read and approved the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors have declared that no competing interest exist.
References
1.
Frade, I.; Ribeiro, A. Bike-sharing stations: A maximal covering location approach. Transp. Res. Part A
2015
,
82, 216–227. [CrossRef]
2.
Ethan, D.; Basch, C.; Johnson, G.; Hammond, R.; Chow, C.; Varsos, V. An analysis of technology-related
distracted biking behaviors and helmet use among cyclists in New York City. J. Community Health
2016
,41,
138–145. [CrossRef] [PubMed]
3.
Schleinitz, K.; Petzoldt, T.; Gehlert, T. Risk compensation? —The relationship between helmet use and
cycling speed under naturalistic conditions. J. Saf. Res. 2018,67, 165–171. [CrossRef] [PubMed]
4.
Fishman, E.; Schepers, P. Global bike share: What the data tells us about road safety. J. Saf. Res.
2016
,56,
41–45. [CrossRef]
5.
Rog
é
, J.; Ndiaye, D.; Vienne, F. Useful visual field training: A way to improve elderly car drivers’ ability to
detect vulnerable road users. Transp. Res. Part F Trac Psychol. Behav. 2014,26, 246–257. [CrossRef]
6.
Twisk, D.A.; Reurings, M. An epidemiological study of the risk of cycling in the dark: the role of visual
perception, conspicuity and alcohol use. Accid. Anal. Prev. 2013,60, 134–140. [CrossRef]
7.
Noordzij, P.C. Cycling in the dark: An analysis of fatal bicycle accidents in The Netherlands. J. Saf. Res.
1976
,
8, 73–76.
8.
Demaio, P. Bike-sharing: History, impacts, models of provision, and future. J. Public Transp.
2014
,12, 3.
[CrossRef]
Sustainability 2019,11, 7035 11 of 11
9.
Midgley, P. Bicycle-Sharing Schemes: Enhancing Sustainable Mobility in Urban Areas. Dep. Econ. Soc. A.
2011,8, 1–12.
10.
Thornley, S.J.; Woodward, A.; Langley, J.D.; Ameratunga, S.N.; Rodgers, A. Conspicuity and bicycle crashes:
Preliminary findings of the taupo bicycle study. Inj. Prev. J. Int. Soc. Child Adolesc. Inj. Prev.
2008
,14, 11.
[CrossRef]
11.
Kwan, I.; Mapstone, J. Visibility aids for pedestrians and cyclists: A systematic review of randomised
controlled trials. Accid. Anal. Prev. 2004,36, 305–312. [CrossRef]
12.
Madsen, J.C.O.; Andersen, T.; Lahrmann, H.S. Safety eects of permanent running lights for bicycles:
A controlled experiment. Accid. Anal. Prev. 2013,50, 820–829. [CrossRef] [PubMed]
13.
Wood, J.M.; Tyrrell, R.A.; Marszalek, R.; Lacherez, P.; Carberry, T.; Chu, B.S. Using reflective clothing to
enhance the conspicuity of bicyclists at night. Accid. Anal. Prev. 2012,45, 726–730. [CrossRef] [PubMed]
14.
Prati, G. The eect of an italian nationwide mandatory visibility aids law for cyclists. J. Transp. Health
2018
,
9, 212–216. [CrossRef]
15.
Lahrmann, H.; Madsen, T.K.O.; Olesen, A.V.; Madsen, J.C.O.; Hels, T. The eect of a yellow bicycle jacket on
cyclist accidents. Saf. Sci. 2017,108, 209–217. [CrossRef]
16.
Carol, C.; Denise, K.; Miller, P.D.; Frank, C. The use of conspicuity aids by cyclists and risk of crashes
involving other road users: A protocol for a population based case-control study. BMC Public Health
2010
,10,
1–6.
17.
National Transport Commission. National Transport Commission Australian Road Rules, Regulation
259. 2009. Available online: http://www.ntc.gov.au/ViewPage.aspx?documentid=00794 (accessed on
20 May 2010).
18.
Wood, J.M.; Tyrrell, R.A.; Marszalek, R.; Lacherez, P.; Carberry, T. Bicyclists overestimate their own night-time
conspicuity and underestimate the benefits of retroreflective markers on the moveable joints.
Accid. Anal. Prev.
2013,55, 48–53. [CrossRef]
19.
Wood, J.M.; Lacherez, P.F.; Marszalek, R.P.; King, M.J. Drivers’ and cyclists’ experiences of sharing the road:
Incidents, attitudes and perceptions of visibility. Accid. Anal. Prev. 2009,41, 772–776. [CrossRef]
20.
Rog
é
, J.; Ndiaye, D.; Aillerie, I.; Aillerie, S.; Navarro, J.; Vienne, F. Mechanisms underlying cognitive
conspicuity in the detection of cyclists by car drivers. Accid. Anal. Prev. 2017,104, 88–95. [CrossRef]
21.
Tin, S.; Woodward, A.; Ameratunga, S. Incidence, risk, and protective factors of bicycle crashes: Findings
from a prospective cohort study in New Zealand. Prev. Med. 2013,57, 152–161. [CrossRef]
22.
Hancock, P.A.; Wulf, G.; Thom, D.; Fassnacht, P. Driver workload during diering driving maneuvers.
Accid. Anal. Prev. 1990,22, 281–290. [CrossRef]
23.
Wulf, G.; Hancock, P.A.; Rahimi, M. Motorcycle conspicuity: An evaluation and synthesis of influential
factors. J. Saf. Res. 1989,20, 153–176. [CrossRef]
24.
Horswill, M.S.; Helman, S.; Ardiles, P.; Wann, J.P. Motorcycle accident risk could be inflated by a time to
arrival illusion. Optom. Vis. Sci. 2005,82, 740–746. [CrossRef] [PubMed]
25.
Mitsopoulos-Rubens, E.; Lenn
é
, M.G. Issues in motorcycle sensory and cognitive conspicuity: The impact of
motorcycle low-beam headlights and riding experience on drivers’ decisions to turn across the path of a
motorcycle. Accid. Anal. Prev. 2012,49, 86–95. [CrossRef] [PubMed]
26.
Prijs, K. Problems of Sensory-and Cognitive Conspicuity of Motorcyclists at Junctions: A Car to Motorcycle
Comparison of Visual Search and Give-Way Intentions by Car Drivers, from two Angles of Approach.
Master’s Thesis, Utrecht University, Utrecht, The Netherlands, 2015.
27.
Clabaux, N.; Brenac, T.; Perrin, C.; Magnin, J.; Canu, B.; Van Elslande, P. Motorcyclists’ speed and
“looked-but-failed-to-see” accidents. Accid. Anal. Prev. 2012,49, 73–77. [CrossRef] [PubMed]
28.
Balk, S.A.; Tyrrell, R.A.; Brooks, J.O.; Carpenter, T.L. Highlighting human form and motion information
enhances the conspicuity of pedestrians at night. Perception 2008,37, 1276–1284. [CrossRef] [PubMed]
29.
Wood, J.M.; Tyrrell, R.A.; Carberry, T.P. Limitations in drivers’ ability to recognize pedestrians at night.
Hum. Factors J. Hum. Factors Ergon. Soc. 2005,47, 644–653. [CrossRef]
©
2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
The promotion of sustainable alternatives to motorized individual mobility has been seen in the past few decades as one of the cornerstones in a strategy to reduce the negative externalities related to the transportation sector. Bicycle sharing is increasingly popular as a sustainable transport system and the number of bike sharing schemes has grown significantly worldwide in recent years. One of the most important elements in implementation of these systems is the location of the stations. In fact the non-optimal locating of bike sharing compromises its success. Municipalities or public–private partnerships are mostly responsible for implementing bike-sharing schemes. The public investment in bicycle mobility (particularly bike-sharing) is complex because it is always subject to a budget. The main concern for public investment is to maximize the benefits through the design and implementation of bike-sharing systems. This work sets out a methodology to help with the decision-making of bike-sharing systems. The research work we present proposes using an optimization method to design the bike sharing system such that it maximizes the demand covered and takes the available budget as a constraint. It combines strategic decisions for locating bike-sharing stations and defining the dimension of the system (stations and number of bicycles) with operational decisions (relocating bicycles). As an outcome, the model determines the optimal location of the bicycle stations, the fleet size, the capacity of the stations and the number of bicycles in each station, considering an initial investment lower than the given budget. In addition, it balances the annual cost of the system and the revenue assuming a possible supplementary budget from the system provider to cover any loss resulting from the shortfall between its operating cost and the revenue from the subscription charges. A case study in Coimbra, Portugal, is presented and discussed.
Article
Full-text available
To describe the methods, characteristics of participants, and report on the preliminary findings of a longitudinal study of cyclists. Web-based survey to establish a cohort of cyclists. Participants in the largest mass-participation bicycle event in New Zealand, the Wattyl Lake Taupo Cycle Challenge. 2469 riders who had enrolled online in the 2006 Wattyl Lake Taupo Cycle Challenge. Self-reported crashes in preceding 12 months. Of 5653 eligible riders, 2469 (44%) completed the study questionnaire. Mean age was 44 years, 73% were male, and the average number of kilometers cycled per week in the preceding 12 months was 130. The annual incidence of crashes leading to injury that disrupted usual daily activities for at least 24 h was 0.5 per cyclist/year. About one-third of these crashes resulted in presentation to a health professional. The mean number of days absent from work attributable to bicycle crashes was 0.39 per cyclist/year. After adjustment for potential confounders and exposure (kilometers cycled per year), the rate of days off work from bicycle crash injury was substantially lower among riders who reported always wearing fluorescent colors (multivariate incidence rate ratio 0.23, 95% CI 0.09 to 0.59). Low cyclist conspicuity may increase the risk of crash-related injury and subsequent time off work. Increased use of high-visibility clothing is a simple intervention that may have a large impact on the safety of cycling.
Article
Introduction: An argument against mandatory helmet use is based on the idea of risk compensation, which means that cyclists might ride faster when wearing a helmet (Lardelli-Claret et al., 2003). However, questionnaire and experimental studies were unable to find evidence for this assumption (Fyhri et al., 2012; Fyhri & Philipps, 2013). Simultaneously, other factors with a potential role in helmet use and cycling speed, such as trip length and rider characteristics have been neglected in such considerations. The goal of the analysis presented in this paper was therefore to investigate the relationship between helmet use and cycling speed under naturalistic conditions while taking characteristics of cyclists and bicycles into account. Method: As part of a naturalistic cycling study, we equipped the bicycles of conventional and e-bike riders with data acquisition systems to record speed and trip distance. It included two cameras (one for the face of the participant, another one for the forward scenery). For the analysis presented in this paper, we used the data of 76 participants (28 conventional bicycles, 48 e-bikes). Results: In total, participants used their helmet for 56% of all trips. Helmets were used more frequently for longer trips. A linear mixed model, in which trip length, helmet use, bicycle type, age, and gender were used as predictors showed that helmet use did not play a significant role for cycling speed. Instead, all other factors that were analysed, with the exception of gender, had a significant relationship to cycling speed. Discussion: The assumption of risk compensation as a result of the use of a helmet could not be confirmed. Instead, the findings seem to support the suggestion that cyclists who undertake trips at potentially higher speed levels are aware of their increased risk, and actively try to reduce it through the use of a helmet.
Article
The role of conspicuity in preventing bicycle–motorized vehicle collisions has been the subject of investigation. To date, no study has evaluated the impact on bicycle safety of legislation imposing bicycling visibility aids. The aim of the present study is to investigate whether a legislation imposing high-visibility clothing for cyclist affects bicycle safety. Data on the monthly number of vehicles (including bicycles) involved in road crashes during the period 2001–2015 were obtained from the Italian National Institute of Statistics. Data were analyzed through an interrupted time-series analysis using a generalized least-squares method. Results revealed that the implementation of legislation imposing high-visibility clothing for cyclist did not influence the number of bicycles involved in road crashes as well as its proportion in the total vehicles involved in road crashes. The introduction of the legislation did not produce immediate effects, nor did it have any effects over time. Lack of knowledge on how the law was introduced, the degree of enforcement by the police, and behavioral changes in response to the law makes it difficult to attribute the lack of effect on bicycle crashes.
Article
This study is the first randomised controlled trial (RCT) of the safety effect of high-visibility bicycle clothing. The hypothesis was that the number of cyclist accidents can be reduced by increasing the visibility of the cyclists. The study design was an RCT with 6793 volunteer cyclists - 3402 test cyclists (with a yellow jacket) and 3391 control cyclists (without the jacket). The safety effect of the jacket was analysed by comparing the number of self-reported accidents for the two groups. The accident rate (AR) (accidents per person month) for personal injury accidents (PIAs) for the test group was 47% lower than that of the control group. For accidents involving cyclists and motor vehicles, it was 55% lower. The study was non-blinded, and the number of reported single accidents was significantly lower in the test group than in the control group. This is likely a result of a response bias, since the bicycle jacket was not expected to affect the number of single accidents. To compensate for this bias, a separate analysis was carried out. This analysis reduced the effect of the jacket from 47% to 38%.
Article
Objective: The aim of this study was to evaluate the visibility of cyclists for motorists in a simulated car driving task. Background: In several cases involving collisions between cars and cyclists, car drivers failed to detect the latter in time to avoid collision because of their low conspicuity. Method: 2 groups of motorists (29.2 years old), including 12 cyclist-motorists and 13 non-cyclist-motorists, performed a vulnerable road user detection task in a car-driving simulator. They had to detect cyclists and pedestrians in an urban setting and evaluate the realism of the cyclists, the traffic, the city, the infrastructure, the car driven and the situations. Cyclists appeared in critical situations derived from previous accounts given by injured cyclists and from cyclists' observations in real-life situations. Cyclist's levels of visibility for car drivers were either high or low in these situations according to the cyclists. Results: Realism scores were similar and high in both groups. Cyclist-motorists had fewer collisions with cyclists and detected cyclists at a greater distance in all situations, irrespective of cyclist visibility. Several mechanisms underlying the cognitive conspicuity of cyclists for car drivers were considered. Conclusion: The attentional selection of a cyclist in the road environment during car driving depends on top-down processing. Application: We consider the practical implications of these results for the safety of vulnerable road users and future directions of research.
Article
Introduction: Bike share has emerged as a rapidly growing mode of transport in over 800 cities globally, up from just a handful in the 1990s. Some analysts had forecast a rise in the number of bicycle crashes after the introduction of bike share, but empirical research on bike share safety is rare. The goal of this study is to examine the impact of bike share programs on cycling safety. Methods: The paper has two substudies. Study 1 was a secondary analysis of longitudinal hospital injury data from the Graves et al. (2014) study. It compared cycling safety in cities that introduced bike share programs with cities that did not. Study 2 combined ridership data with crash data of selected North American and European cities to compare bike share users to other cyclists. Results: Study 1 indicated that the introduction of a bike share system was associated with a reduction in cycling injury risk. Study 2 found that bike share users were less likely than other cyclists to sustain fatal or severe injuries. Conclusions: On a per kilometer basis, bike share is associated with decreased risk of both fatal and non-fatal bicycle crashes when compared to private bike riding. Practical applications: The results of this study suggest that concerns of decreased levels of cycling safety are unjustified and should not prevent decision makers from introducing public bike share schemes, especially if combined with other safety measures like traffic calming.
Article
Bicycling is becoming an increasingly utilized mode of transportation in New York City. Technology-related distracted bicycling and helmet use are behaviors that can impact bike safety. The aims of this study were twofold: (1) to determine rates and types of technology-related distracted behaviors among bicyclists in the borough of Manhattan in New York City; and (2) to assess the rate of bicycle helmet use among these cyclists. Bicyclists in five popular riding areas in Manhattan were observed for a total of 50 h using a digital video camera during summer months in 2014. Videos were coded and enumerated for the total number and gender of cyclists, type of bicycle, number wearing headphones/earbuds and/or using a mobile phone, and whether the cyclist was wearing a helmet. Almost 25,000 cyclists were observed across the five selected locations (n = 24,861). Riders were almost four times more likely not to wear a helmet on rental bikes as compared with non-rentals (Citi Bike(®) OR 3.8; 95 % CI 2.5, 5.9: other rental OR 3.8; 95 % CI 3.0, 4.9). Significantly increased odds of not wearing a helmet were observed for females relative to males (OR 1.4; 95 % CI 1.1, 1.8) across varied times and locations. Overall, rates of technology-related distraction were low, with headphone use being most prevalent. Males were more likely to wear headphones/earbuds (OR 2.0; 95 % CI 1.4, 2.9), as were cyclists on Citi Bikes relative to other rental bikes (OR 2.2; 95 % CI 1.3, 3.6). Findings from this study contribute to the growing literature on distracted biking and helmet use among bike share program riders and other cyclists and can inform policymakers and program planners aiming to improve bicycle safety in urban settings.
Article
This paper presents statistical data on the proportions of daytime and dusk/nighttime cycling in urban and rural areas and the respective proportions of accidents. Survey data show that the average percentage of encounters between cyclists and traveling motor vehicles throughout the year is 92% during the day and 8% at night. However, the risk of having a fatal collision is nearly 4 times greater under conditions of darkness. In addition, the risk of a fatal impact was shown to be particularly great for head-tail collisions in rural areas at night.
Article
Our study examined the ability of two groups of elderly car drivers to detect vulnerable road users (pedestrians and two-wheeled motorized vehicles) in a simulated car-driving task. One group was given specific training, aimed at increasing their useful visual field, while driving. The second group’s task was to follow another vehicle. Data analysis indicated that the training given led to a significant increase in drivers’ useful visual field. In addition, comparison of performances of the two groups permitted us to conclude that training had a beneficial effect on driving, as elderly drivers who received training were better able to detect pedestrians in the road environment. This research could potentially be applied in cognitive intervention programs for older adults in order to preserve their ability to drive and improve the safety of vulnerable road users.