International Journal of

Environmental Research
and Public Health

Article
Effect of Using Mobile Phones on Driver’s Control
Behavior Based on Naturalistic Driving Data
Lanfang Zhang 1 , Boyu Cui 1 , Minhao Yang 2 , Feng Guo 3 and Junhua Wang 1, *
1 Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University,
Shanghai 201804, China; zlf2276@tongji.edu.cn (L.Z.); 1733299@tongji.edu.cn (B.C.)
2 Shanghai Urban Construction Design and Research Institute, Shanghai 200125, China;
yangminhao@sucdri.com
3 Department of Statistics, Virginia Polytechnic and State University, Blacksburg, VA 24061, USA;
feng.guo@vt.edu
* Correspondence: benwjh@163.com; Tel.: +86-130-6176-5132




Received: 2 April 2019; Accepted: 23 April 2019; Published: 25 April 2019


Abstract: Distracted driving behaviors are closely related to crash risk, with the use of mobile phones
during driving being one of the leading causes of accidents. This paper attempts to investigate the
impact of cell phone use while driving on drivers’ control behaviors. Given the limitation of driving
simulators in an unnatural setting, a sample of 134 cases related to cell phone use during driving
were extracted from Shanghai naturalistic driving study data, which provided massive unobtrusive
data to observe actual driving process. The process of using mobile phones was categorized into five
operations, including dialing, answering, talking and listening, hanging up, and viewing information.
Based on the concept of moving time window, the variation of the intensity of control activity, the
sensitivity of control operation, and the stability of control state in each operation were analyzed. The
empirical results show strong correlation between distracted operations and driving control behavior.
The findings contribute to a better understanding of drivers’ natural behavior changes with using
mobiles, and can provide useful information for transport safety management.

Keywords: natural driving data; distracted driving; mobile phone; moving time window; driving
control behavior

1. Introduction
Related studies have shown that using mobile phones during driving is one of the leading causes
of traffic accidents [1–8]. In response to this problem, nearly 70 countries and regions have enacted laws
and regulations banning the use of mobile phones during driving [9]. However, most countries only
prohibit the making or receiving of hand-held calls, and there is no prohibition on the use of hands-free
devices [10]. Differences in regulations indicate that there is still a lack of unified understanding of the
impact of using mobile phones on driving behavior. At the same time, these regulations haven’t been
widely accepted by drivers [11–14], and this distracted behavior is becoming more frequent with the
explosive development of smartphone functions.
Driving tasks mainly include controlling the stability of vehicles and monitoring the driving
environment, which are directly related to driver’s manual operation and visual attention. Some
previous studies tend to take visual distraction behavior as the main factor of accident. For example, the
mean percentage of “total eyes-off-road time” (TEORT) is 33.1% and 59.5% when the driver performs
answering and dialing operations on a hand-held call. However, the percentage of TEORT is 9.5%
or 15.6% when the driver communicates by phone in the hand-held or hands-free mode [15,16]. In
addition, most operations of using mobile phones are manual–visual distractions, such as texting and

Int. J. Environ. Res. Public Health 2019, 16, 1464; doi:10.3390/ijerph16081464 www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2019, 16, 1464 2 of 13

dialing. The effect of manual–visual distractions on driving performance is negative and of greater
impact than just visual distraction [17–19]. Therefore, many countries or regions, when formulating
relevant laws and regulations, believe that manual–visual interference in handheld phone calls will
have a significant impact on driving safety, while ignoring the impact of cognitive interference on
driving behavior. However, the cognitive distraction caused by using mobile phones will affect the
performance of driving tasks from two aspects according to ergonomics theory. Cognitive psychology
holds that driver’s attention during driving is limited [20]. Driver’s attention typically splits between
performing control operations and processing driving environment information. When the distraction
demand caused by mobile phone use is high, a part of attention is required to be allocated to deal with
the distracted operation, thus causing interference to the driver’s control behavior and environmental
monitoring ability. According to Wickens’ Multiple Resource theory [21], although conversation is
auditory and driving tasks are visual, cognitive distraction interferes with visual behavior as the
conversation triggers driver’s visual memory. As a result, drivers’ ability to process visual information
becomes more sluggish, and less sensitive to performing control operations and monitoring the driving
environment. Collet’s study compared changes in driver’s heart rates and skin resistance during
hand-held calls and conversations with passengers. The results show that when drivers engage in
these two kinds of distracted behaviors, their physiological indicators are significantly changed and
there is no significant difference as to the degree of influence [22].
Early studies on distracted driving related to mobile phone use were mainly based on using
driving simulators to explore the impact on driving behavior. It was found that hand-held calls or
sending and receiving texting messages during driving caused a significant decrease in speed, making
it difficult to maintain stable speed and headway distance [23–25]. In the lateral direction, the lateral
deviation increased, the frequency of lane changes and lateral stability decreased [26,27]. At the same
time, drivers’ ability to recognize signals was compromised and their reaction time lengthened [28–30].
The driving simulation environment provides the opportunity of data acquisition for related research,
and has achieved a lot of research results. However, the driving simulation research is experimental
and participants are often required to perform distraction behavior in a relatively unfamiliar driving
environment. Participants couldn’t choose whether to perform distracted behaviors according to
the driving state and environment, which is quite different from real situations and leads to certain
limitations of results.
With the launch of large-scale natural driving projects, such as the Second Strategic
Highwasy Research Plan Naturalistic Driving Study (SHRP2 NDS), UTDrive, and Shanghai NDS
(SH-NDS), researchers began to conduct mobile phone distraction study based on natural driving
data [7,8,10,15,31–34]. Natural driving research reduces the interference of experimental arrangements
on driving behavior and provides the opportunity to observe the actual driving process with unobtrusive
high precision data [35]. However, unlike a lab-controlled experiment, many more factors of the
driving environment should be taken into consideration. Most research didn’t consider the interference
of driving factors such as road scenes and traffic conditions, which affected the results. In addition, the
above studies generally use the mean or instantaneous values of parameters such as speed, distance,
lane offset or steering wheel angle when constructing indexes to represent the running state of vehicles.
In the state of natural driving, the driver will make some self-adjustment when the distraction is
weak, so that these indicators may not be significantly affected, and it is difficult to fully represent the
performance of the driver and changes of driving state in the whole process. In addition, the process
of using mobile phones for hand-held calls involves multiple operations such as answering, dialing,
talking, and hanging up. The distractions caused by different operations differ in their type, intensity
and duration. In previous studies, the influence of mobile phone use on driving performance and
hand-held call behavior is often regarded as a whole process without distinguishing and comparing
different operations according to the characteristic of distraction. In summary, the understanding of
the influence of using mobile phones on driving behaviors, especially drivers’ control behaviors, under
natural driving conditions is still vague.
Int. J. Environ. Res. Public Health 2019, 16, 1464 3 of 13

Based on the natural driving data of Shanghai NDS, this paper attempts to explore the influence
Int. J. Environ. Res. Public Health 2019, 16, x 3 of 13
on driver’s control behavior of five operating behaviors of using mobile phones in a natural driving
state:Based
answering,
on the dialing, talking data
natural driving and listening,
of Shanghaihanging up and
NDS, this viewing
paper attemptsinformation. We influence
to explore the collected
samples based on pre-defined screening conditions to reduce the influence of driving
on driver’s control behavior of five operating behaviors of using mobile phones in a natural driving environment
factors, and categorize
state: answering, the talking
dialing, different operations
and listening,ofhanging
mobile phone
up and use into five
viewing types. Based
information. Weoncollected
driver’s
distraction
samples based mechanism, multiple
on pre-defined metrics conditions
screening representing to the driver’s
reduce control behavior
the influence areenvironment
of driving constructed
using
factors, and categorize the different operations of mobile phone use into five types. Basedoperation’s
moving time window, including the index of control activity’s intensity, control on driver’s
sensitivity, and control state’s stability.
distraction mechanism, multiple metrics representing the driver’s control behavior are constructed
using
2. moving time window, including the index of control activity’s intensity, control operation’s
Methods
sensitivity, and control state’s stability.
2.1. Shanghai Natural Driving Study
2. Methods
The Shanghai Natural Driving Research Project (SH-NDS), jointly conducted by the Tongji
university,
2.1. ShanghaitheNatural
General Motors,
Driving and Virginia Tech Transportation Institute, is the first study on natural
Study
driving in China. The project collected the driver’s behavior for an extended time period (up to
The Shanghai
two months), Natural Driving
thus minimizing Research Project
the interference to the (SH-NDS), jointly
driver’s daily conducted
driving by the
behavior. Tongji
A total of
university, the General Motors, and Virginia Tech Transportation Institute, is
60 drivers, aged between 35 and 50, participated in the study, and each driver had more than fivethe first study on
natural
years of driving in China.
prior driving The project
experience. Thecollected the driver’s
experimental vehiclesbehavior
includedfor anGM
five extended
brandedtime periodwith
vehicles (up
to two months), thus minimizing the interference to the driver’s daily driving behavior.
automatic transmission vehicles. To date, six phases of the project have been launched, with more A total of 60
drivers,
than aged km
750,000 between 35 anddata
of driving 50, participated
collected. Asinshown
the study, and each
in Table 1 anddriver
Figure had
1, more thanacquisition
the data five years
of prior driving experience. The experimental vehicles included five GM branded
system (DAS) consists of Doppler radar, Triaxial accelerometer, GPS and four synchronous cameras. vehicles with
automatic transmission vehicles. To date, six phases of the project have been launched,
The DAS can continuously collect driving state information, driver behavior information and external with more
than 750,000 km
environment of driving
information data collected. The
simultaneously. As shown
stored in Table
data can 1beand Figure
divided 1, the
into data acquisition
numerical data and
system (DAS)
video data. consists of Doppler radar, Triaxial accelerometer, GPS and four synchronous cameras.
The DAS can continuously collect driving state information, driver behavior information and external
environment information simultaneously.
Table The stored
1. Function introduction dataacquisition
of data can be divided
devices.into numerical data and
video data.
Data Acquisition Equipment Introduction

Table 1. FunctionThe equipment of

introduction collects
data the relative distance
acquisition devices.and relative speed with
Doppler radar surrounding vehicles. The measuring range is 40 m transversal and
Data Acquisition 150 m longitudinal. The acquisition frequency is 10 Hz.
Introduction
Equipment
The acceleration and angular velocity of the vehicle in three
The equipment collects the relative distance and relative speed with surrounding vehicles. The
Triaxial
Doppler radaraccelerometer directions are collected for determining the vehicle motion state, and
measuring range is 40 m transversal and 150 m longitudinal. The acquisition frequency is 10 Hz.
the data acquisition frequency is 10 Hz.
The acceleration and angular velocity of the vehicle in three directions are collected for
Triaxial accelerometer
determining theVehicle
vehicle coordinate
motion state,information
and the data is collected,frequency
acquisition and dataisacquisition
10 Hz.
Global Positioning System (GPS)
Global Positioning frequency is 1 Hz.
Vehicle coordinate information is collected, and data acquisition frequency is 1 Hz.
System (GPS)
A total of four cameras capture the driving behavior of the vehicle’s
Multi-cameraA total of four cameras capture the driving behavior of the vehicle’s front view, rear view,
Multi-camera front view, rear view, driver’s face and hand.
driver’s face and hand.

Figure
Figure1.1.The
The framework
framework ofof video
video data
data in SH-NDS
in SH-NDS (Shanghai
(Shanghai NaturalNatural Driving Project).
Driving Research Research
Project).

2.2. Sampling
Under natural driving conditions, changes in road environment and traffic conditions have a
great impact on driving control behavior. In order to reduce the external factors that may interfere
Int. J. Environ. Res. Public Health 2019, 16, 1464 4 of 13

2.2. Sampling
Under natural driving conditions, changes in road environment and traffic conditions have a
great impact on driving control behavior. In order to reduce the external factors that may interfere
with the analysis results, five sample screening conditions were set in the sample acquisition process,
listed below.

• The vehicle was driving on an expressway or freeway when the event occurred.;
• The vehicle was in continuous traffic flow.
• No other events occurred at the same time.
• No obvious interference from surrounding vehicles (no emergency brake or sharp turn)
was detected.
• There was a period of normal driving before and after the event to get control samples.

The distracted driving sample was screened by checking the driver’s facial video and hand video.
The beginning of a cellphone use event is defined as when the driver’s eye-glance shifted and a hand
began to reach for the phone. The event is considered to be over when the driver’s eye-glance returned
to driving tasks and hands were back on the steering wheel. To meet the screening conditions, the
surrounding driving environment condition for each cellphone use event was examined by checking
the front video and the rear video. Moreover, video data and numerical data in the Shanghai natural
driving project adopt a unified timeline, and the corresponding bus data can be extracted through the
timestamp in the video.

2.3. Cellphone Subtasks

The use of mobile phones during driving can be classified into two types: the hand-held call
behavior and the information viewing behavior. Among them, hand-held call behavior is further
categorized into answering calls and making calls. As shown in Table 2, the hand-held call behavior
includes four sub-tasks: answering or dialing, talking and listening, and hanging up. Amongst
these sub-tasks, the visual distraction is reduced subsequently into viewing, dialing, answering and
hanging up. In addition, answering, dialing and viewing create conspicuous manual distraction. The
communication process mainly leads to cognitive distraction of the driver. Table 2 lists the criteria
for the occurrence of each operation, and the occurrence of the next operation or normal process
determined the end of the previous operation. This paper obtained the numerical driving data in the
corresponding operation process by recording the time stamp in the video when the operation occurs
and ends. In addition, as shown in Figure 2, the normal driving process of 3 s before and after the use
of mobile phone was taken as the control case. At the same time, in order to avoid the influence of
using mobile phone on the control sample, a 1 s interval was set between the normal process interval
and the distraction interval.

Table 2. Description of distraction operations.

Task Sub-Operations Description Judging Criteria

Check the incoming call with one Driver’s line of sight starts
Answering
hand and prepare the call. to move.
Click the screen with one Driver’s line of sight starts
Dialing
hand to dial. to move.
Hand-held call
Drive with one hand and
Talking and listening Put the phone at the ear.
communicate with a phone.
Stop talking and put the phone Put the phone away from
Hanging up
down with one hand. the ear.
Click on the screen with one hand to
Driver’s line of sight starts
View information Viewing browse information. Keep eyes off
to move.
the road for a long time.
Int. J. Environ. Res. Public Health 2019, 16, 1464 5 of 13
Int. J. Environ. Res. Public Health 2019, 16, x 5 of 13

Figure
Figure2.2.The
Thearea
area of data
dataextraction.
extraction.

2.4. Data Analysis

2.4. Data Analysis
In the process of using mobile phones, drivers will reallocate their attention resources to cope with
In the process of using mobile phones, drivers will reallocate their attention resources to cope
the workload of distracted operations on driving tasks, which will affect the intensity of relevant control
with the workload of distracted operations on driving tasks, which will affect the intensity of relevant
activities. At the same time, distracted operation will make the driver more sluggish in acquiring and
control activities. At the same time, distracted operation will make the driver more sluggish in
processing environmental information, and thus the sensitivity of driving control operation will be
acquiring and processing environmental information, and thus the sensitivity of driving control
partially compromised. Under the influence of the above two effects, the stability of driving control
operation will be partially compromised. Under the influence of the above two effects, the stability
state will likely to fluctuate accordingly. Therefore, this paper examines the driving control behavior
of driving control state will likely to fluctuate accordingly. Therefore, this paper examines the driving
based on the dynamic data collected in the natural driving project, and analyzes the changes of the
control behavior based on the dynamic data collected in the natural driving project, and analyzes the
intensity
changes of of the
driver’s control
intensity of activities, the sensitivity
driver’s control of the
activities, control operation,
sensitivity and theoperation,
of control stability ofandcontrol
the
state under
stability different
of control operations.
state under different In theoperations.
process of In using mobile phones,
the process of using there
mobile is phones,
a high degree
there isofa
distraction
high degreeinofspecific moments,
distraction in specific which has a significant
moments, which hasimpact on theimpact
a significant driver’soncontrol behavior.
the driver’s controlIn
order to explore
behavior. In order thistoeffect,
explore it isthis
necessary
effect, ittoismake full use
necessary to of the sequence
make full use ofdata theof the natural
sequence datadriving
of the
study to construct indexes that can obtain the running characteristics of
natural driving study to construct indexes that can obtain the running characteristics of vehicles vehicles in the process of using in
mobile phones.
the process of using mobile phones.
In this
In this paper,
paper, the
the concept
concept of of moving
moving time window was
time window was used
used to to construct
construct thethe characteristic
characteristic index
index
of
of driver’s control behavior [36]. As shown in Figure 3 α represents the parameter value, such as
driver’s control behavior [36]. As shown in Figure 3 α represents the parameter value, such as
velocity
velocity and acceleration, and the standard deviation of α was used as a measure of fluctuation in as
and acceleration, and the standard deviation of α was used as a measure of fluctuation in a 1
time window
1 s time windowto detect the minimum
to detect the minimum variation of the parameter.
variation The collection
of the parameter. frequency
The collection of dynamic
frequency of
data
dynamic data in SH-NDS is generally 10 Hz, so a 1 s time window can be built every 0.1 s. In of
in SH-NDS is generally 10 Hz, so a 1 s time window can be built every 0.1 s. In the process thea
sub-operation, the standard deviation
process of a sub-operation, the standard of parameters
deviation of within each 1 swithin
parameters time window
each 1 swas time calculated,
window was and
the standard deviation of the i th window is denoted by STD , forming a new sequence composed of
calculated, and the standard deviation of the 𝑖 window is denoted by 𝑆𝑇𝐷 , forming a new sequence
i
standard
composeddeviations.
of standard deviations.
In the
In the state
state of
of natural
natural driving,
driving, vehicle
vehicle kinematic
kinematic parameters
parameters such such as as speed,
speed, acceleration
acceleration and and lane
lane
deviation are sensitive to road traffic condition and the driver’s own
deviation are sensitive to road traffic condition and the driver’s own driving intention. However, driving intention. However,
when
when thethe driver’s control ability
driver’s control ability remains
remains in in aa normal
normal state
state and
and isis not
not strongly
strongly affected
affected by by the driving
the driving
environment (such as sharp turning, traffic congestion, etc.), the
environment (such as sharp turning, traffic congestion, etc.), the fluctuation of these kinematic fluctuation of these kinematic
parameters
parameters within
withineacheachsecond
secondwill willlikely
likelytotoremain
remain within
within a certain range
a certain without
range withoutdrastic changes.
drastic The
changes.
fluctuation
The fluctuationof parameter
of parametervaluesvalues
will bewill
smaller in a short
be smaller inperiod
a short under
period normal
under conditions. Therefore,
normal conditions.
under the premise of no drastic changes under normal driving environment,
Therefore, under the premise of no drastic changes under normal driving environment, this stability this stability may be lost
when the driver’s control ability is affected significantly. Therefore, the
may be lost when the driver’s control ability is affected significantly. Therefore, the one-second one-second standard deviation
value
standardcalculated
deviation in each
valuetime window
calculated in reduces
each time thewindow
interference
reduces of other factors andof
the interference can be used
other factors to
characterize the influence of distracted operation. In addition, the influence
and can be used to characterize the influence of distracted operation. In addition, the influence of of distracted operation
will vary with
distracted the degree
operation will of distraction
vary with theindegree
the process of using ainmobile
of distraction phone.ofThis
the process using processing
a mobilemethod
phone.
can
This processing method can make full use of the sequence data for preliminary featureofextraction.
make full use of the sequence data for preliminary feature extraction. The sequence standard
deviation
The sequence valuesof obtained
standard by preliminary
deviation values calculation
obtained can ensure that sufficient
by preliminary calculationdata canhas a strong
ensure that
correlation with the distracted operation and facilitate the subsequent construction
sufficient data has a strong correlation with the distracted operation and facilitate the subsequent of feature indicators.
construction of feature indicators.
Int. J. Environ. Res. Public Health 2019, 16, 1464 6 of 13
Int. J. Environ. Res. Public Health 2019, 16, x 6 of 13

Figure3.
Figure Moving time
3. Moving timewindow.
window.
2.4.1. Control Activity Intensity
2.4.1. Control Activity Intensity
The intensity of the control activity reflects the frequency and enthusiasm of the driver in the
The intensity of the control activity reflects the frequency and enthusiasm of the driver in the
control operation and represents the attention resource allocation of the driver in the execution of
control operation and represents the attention resource allocation of the driver in the execution of the
the primary driving tasks. The acceleration of a vehicle is a parameter directly related to driving
primary driving tasks. The acceleration of a vehicle is a parameter directly related to driving control
control operations, such as controlling accelerator or brake pedals, the steering wheel, etc. Therefore,
operations, such as controlling accelerator or brake pedals, the steering wheel, etc. Therefore,
longitudinal acceleration and lateral acceleration are adopted in this paper to construct characteristic
longitudinal acceleration and lateral acceleration are adopted in this paper to construct characteristic
indexes representing the driver’s control activities. The standard deviation of the acceleration value in
indexes representing the driver’s control activities. The standard deviation of the acceleration value
a 1 s window reflects the strength of the driver’s control activities in that second. In this paper, the
in a 1 s window reflects the strength of the driver’s control activities in that second. In this paper, the
median value of the standard deviation sequence of the acceleration is selected to represent the control
median value of the standard deviation sequence of the acceleration is selected to represent the
activity of the driver in each process.
control activity of the driver in each process.
2.4.2. Control Operation Sensitivity
2.4.2. Control Operation Sensitivity
The sensitivity of the control operation is mainly related to the driver’s awareness of road
The sensitivity
information and the of the control
execution of theoperation is mainly related
control operation. The change to the
of driver’s awarenessneeds
control operation of roadto
information and the execution of the control operation. The change
correspond with the change of driving environment in time. When the degree of driver distraction of control operation needs to
correspond with the change of driving environment in time. When the degree
is high, the sensitivity of executing control operation begins to decline. In this paper, the standard of driver distraction is
high, the sensitivity of executing control operation begins to decline. In
deviation sequence of lateral and longitudinal acceleration values during each operation was calculated. this paper, the standard
deviation
In sequence
the Ith time window, ofthelateral
STDi and
valueslongitudinal acceleration
of the longitudinal valuesaccelerations
and lateral during each operation
were was
constructed
calculated. In the 𝐼 time window, the 𝑆𝑇𝐷 values of the longitudinal and
as two-dimensional vectors to represent the overall control activities of the driver in the i window. If lateral accelerations
th were
constructed
the as two-dimensional
driver’s cognitive delay occurs, vectors to represent
the time to changethe from overall controlactivity
one control activities of the
state driver in
to another the
state
𝑖 window.
will increase,Ifwhich
the driver’s cognitive
is manifested as adelay occurs,decrease
significant the timein tothe
change from one
sensitivity control
of the activity
control state
operation
to another state will increase, which is manifested as a significant decrease
within a certain time range. Therefore, the standard Euclidean distance (d) of vectors of adjacent in the sensitivity of the
control operation
windows was further within a certain
calculated time range.
to represent Therefore,
the change of thethe standard
driver’s Euclidean
control activity distance (d) of
at the adjacent
vectors of adjacent windows was further calculated to represent the
moment, and the median value of all the distance values in a process was used to characterize the change of the driver’s control
activity at the
sensitivity adjacent
of the driver’smoment,
controland the median
operation change value of all
in that the distance values in a process was used
process.
to characterize the sensitivity of the driver’s control operation change in that process.
v
u
u  2
u 2 SD𝑆𝐷(i),j ,− SD 𝑆𝐷
tX
(i+1),j ,
d𝑑 = (1)
(1)
S j𝑆2
j=1

where: 𝑖 is the sequence number of the moving time window; the value of j is {1,2}, representing two
where: i is the sequence number of the moving time window; the value of j is {1,2}, representing two
types of acceleration; 𝑆𝐷 is the standard deviation of the j acceleration value in the 𝑖 window; Sj
types of acceleration; SD(i),j, is the standard deviation of the j acceleration value in the ith window; Sj
representsthe
represents thestandard
standard deviation
deviation of the
of the sequence
sequence of standard
of the the standard deviation
deviation of the of the j acceleration
j acceleration during
during the whole process; d is the calculated value of the standard Euclidean distance of the adjacent
the whole process; d is the calculated value of the standard Euclidean distance of the adjacent points.
points.
2.4.3. Control State Stability
2.4.3. Control State Stability
Speed and lane offset are direct driving performance of vehicles in two directions, which have a
strongSpeed and lane
correlation offset
with are direct
driving safety.driving performance
Therefore, this paperof used
vehicles
the in two directions,
values of speed andwhich
lanehave
offseta
strong correlation with driving safety. Therefore, this paper used the values of speed and
to construct indexes to represent the stability of longitudinal and lateral control state of drivers tolane offset to
construct indexes to represent the stability of longitudinal and lateral control state of drivers to evaluate
evaluate the stability of control states. As shown in Figure 4, S j is the standard deviation of the standard
the stability of control states. As shown in Figure 4, 𝑆 is the standard deviation of the standard
deviation sequence of speed values during the whole process, which was used to characterize the
fluctuation of the longitudinal driving control state of the process. Similarly, the 𝑆 value of lane offset
Int. J. Environ. Res. Public Health 2019, 16, 1464 7 of 13

deviation
Int. J. Environ.sequence of speed
Res. Public Health 2019,values
16, x during the whole process, which was used to characterize the
7 of 13
fluctuation of the longitudinal driving control state of the process. Similarly, the S j value of lane offset
was usedused toto represent
represent thethe stability
stability of
of the
thedriver’s
driver’s lateral
lateral control
control state
state during
during the theoperation. The S𝑆j
operation. The
represents the fluctuation of the control state of a certain parameter in a process, not only the fluctuation
of the parameter, so it can can bebe used
used totoevaluate
evaluatethe
thestability
stabilityofofthe
thecontrol
controlstate
stateofofthe
thewhole
wholeprocess.
process.

Figure 4. Calculation
Figure diagramofof𝑆S.j .
diagram
4. Calculation

3. Results
3. Results
As
As shown
shown in Table 3,
in Table 3, according
according to
to the
the sample
sample screening
screening conditions,
conditions, aa sample
sample of
of 134
134 cases
cases of
of five
five
kinds of distracted operations involving 25 drivers are extracted in this paper to analyze the influence
kinds of distracted operations involving 25 drivers are extracted in this paper to analyze the influence
of
of different
different operations
operations on
on driving
driving control
control behaviors.
behaviors.

Table 3. Sample description of each operation.

Table 3. Sample description of each operation.
Sample SizeSize
Sample
Task Category
Task Category(Number of Events) Sub-Operation
Sub-Operation Average
Average Duration
Duration (s) (s)
(Number of events)
Answering
Answering 5.5 5.5
Answer theAnswer
call the call60 60 Talking
Talkingand
and listening
listening 50.1 50.1
Hanging up 5.4
Hanging up 5.4
Dial the call 39 Dialing 21.5
Dial the call
View information 35 39 Dialing
viewing 21.5 25
View information 35 viewing 25

3.1. Answering the Call

3.1. Answering the Call
During the answering process, the intensity of the driver’s longitudinal control activity was
During
decreased the answering
significantly process,
(T(59) = 2.659;thep intensity
= 0.010), ofbutthe
nodriver’s longitudinal
significant control
changes were activity
found was
in other
decreased significantly (T(59) = 2.659; p = 0.010), but no significant changes were
sub-operations. As shown in Table 4, Figures 5 and 6, the sub-operations of answering, talking andfound in other sub-
operations.and
listening, Ashanging
shown in upTable
all led4,toFigures 5 and decrease
a significant 6, the sub-operations of answering,
in the sensitivity talking
of the driver’s and
control
listening, and hanging up all led to a significant decrease in the sensitivity of the driver’s
operation. In the process of answering and talking, the stability of the longitudinal control state was control
operation. In affected,
significantly the process
andofthe
answering
influenceand talking,and
of talking the listening
stability of thesignificantly
was longitudinalhigher
controlthan
stateother
was
significantly
distracted affected, In
operations. and the influence
addition, of talking
the statistical andshow
results listening
that was
theresignificantly higherinfluence
was no significant than otherof
distracted operations. In addition, the statistical results show that there was no significant
the call answering behavior on the driver’s lateral control activities (F(1.6, 94.8) = 3.033, p = 0.064) influence
and
of the call
lateral answering
control behavior
state (F(2.9, 78.6) on the driver’s
= 0.668, lateral control activities (F(1.6, 94.8) = 3.033, p = 0.064)
p = 0.570)).
and lateral control state (F(2.9, 78.6) = 0.668, p = 0.570)).
Table 4. Results summary of main effects and comparisons during answering calls.
Table 4. Results summary of main effects and comparisons during answering calls.
Comparison Control Sensitivity Control State
Comparison Control Sensitivity Control State
A1 –A2 A1–A2 T(59) = 3.010,
T(59) p=
= 3.010, p =0.004
0.004 p ==0.016
T(59)
T(59) = −2.484, −2.484, p = 0.016
A1 –A3 A1–A3 T(59) = p =
T(59) = 3.580, p =0.001
3.580, 0.001 T(59) = −5.751, p < 0.01
T(59) = −5.751, p < 0.01
A2 –A3 A2–A3 T(59) = 0.915,
T(59) p=
= 0.915, p =0.364
0.364 p ==0.008
T(59)
T(59) = −2.752, −2.752, p = 0.008
A3 –A4 A3–A4 T(59) = −0.860,
T(59) = −0.860,p p==0.393
0.393 T(59) = 6.446, p < 0.01
T(59) = 6.446, p < 0.01
A3 –A5 A3–A5 T(59) = −2.541,
T(59) = −2.541,p p==0.014
0.014 T(59) = 5.420, p < 0.01
T(59) = 5.420, p < 0.01
A4 –A5 A4–A5 T(59) = −2.528,
T(59) = −2.528,p p==0.014
0.014 T(59) = −0.165, p = 0.869
T(59) = −0.165, p = 0.869
 Int. J. Environ. Res. Public Health 2019, 16, x 8 of 13
Int. J. Environ. Res. Public Health 2019, 16, 1464 8 of 13
Int. J. Environ. Res. Public Health 2019, 16, x 8 of 13

Figure 5. Statistical results of sensitivity index during answering calls.

Figure
Figure 5. 5. Statisticalresults
Statistical results of
of sensitivity
sensitivityindex
indexduring
duringanswering calls.calls.
answering

Figure
Figure 6. Statisticalresults
6. Statistical resultsof of longitudinal
longitudinal stability
stability index
index during
duringanswering
answering calls.calls.

3.2. DialingFigure
the Call 6. Statistical results of longitudinal stability index during answering calls.
3.2. Dialing the Call
The statistical
3.2. Dialing the Callresults show that compared with the normal driving process, the dialing operation
The statistical results show that compared with the normal driving process, the dialing operation
resulted in a significant decline of driver’s longitudinal control activity (T(38) = 2.045, p = 0.048), a
resulted
The in a significant decline
statistical of driver’s longitudinal controldriving
activityprocess,
(T(38) = 2.045, p =operation
0.048), a
significant decline inresults show that
the sensitivity compared
of the control with the normal
operation (T(38) = 2.315, p = 0.026),the and dialing
a deterioration
significant decline in the sensitivity of the control operationcontrol(T(38) =activity
2.315, p(T(38)
= 0.026), and a deterioration
in the stability of the longitudinal control state (T(38) = −4.466, p < 0.01). However, no significanta
resulted in a significant decline of driver’s longitudinal = 2.045, p = 0.048),
in the stability
significant of the
in longitudinal control state (T(38) = −4.466, p < 0.01). However, noandsignificant change
change wasdecline
found in the
the sensitivity of the
lateral control control
behavior operation
(T(38) = 1.385, (T(38) p ==2.315,
0.174;pT(38)
= 0.026),=−0.806, a deterioration
p = 0.425).
was
in thefound in the
stability lateral
of the control behavior
longitudinal control state (T(38)(T(38)
= 1.385,
= −4.466, p < 0.01).
p = 0.174; T(38) However,
=−0.806, p no = 0.425).
significant change
wasViewing
3.3. found in the lateral control behavior (T(38) = 1.385, p = 0.174; T(38) =−0.806, p = 0.425).
Information
3.3. Viewing Information
In terms of control activities, the longitudinal control activity in the process of viewing information
3.3. Viewing
In affected
terms Information
of control activities, theT(34)longitudinal
weren’t (T(34) = 1.839, p = 0.075; = −0.617, control
p = 0.514), activity
but theinintensity
the process of viewing
of lateral control
information
In was
activity terms weren’t
of control
significantly affected (T(34)(T(34)
activities,
decreased =the1.839, p = 0.075;
longitudinal
= 2.498, T(34)
control
p = 0.017; = −0.617,
T(34) activity p =in0.514),
= −2.044, pthe but the of
process
= 0.049). intensity
viewing of
lateral control
information activity
weren’t was
affected significantly
(T(34) = decreased
1.839, p = (T(34)
0.075; =
T(34) 2.498,
= p
−0.617, =
In terms of control sensitivity, driver’s sensitivity of control operation in the process of viewing0.017;
p = T(34)
0.514), =
but−2.044,
the p = 0.049).
intensity of
was Incontrol
lateral terms of
significantly control
activity
lowerwas sensitivity,
than driver’s
significantly
that before sensitivity
decreased
distraction. (T(34) of =control
The 2.498, operation
results pshow
= 0.017; in there
T(34)
that the=process
−2.044, pof=viewing
0.049).
were significant
was significantly
In terms of lower
control than that
sensitivity, before
driver’s distraction.
sensitivity The
of results
control show
operation
differences between the control samples before and after the distraction. Therefore, this paper further that
in there
the were
process significant
of viewing
differences
was
selected fourbetween
significantly
segments theofcontrol
lower than
2–5 sthatsamples
(N beforebefore and after
distraction. Thetheresults
distraction.
showTherefore,
that therethis paper
were further
significant
3(2s) ), 3–6 s (N3(3s) ), 4–7 s (N3(4s) ) and 5–8 s (N3(5s) ) after the end of
selected four
differencesas
distraction segments
between
control the of 2–5
control
samples s (N
forsamples
3(2s) ),
comparison. 3–6 s
beforeAs(N
and
3(3s)
shown ), 4–7
afterin s
the (N ) and
distraction.
Figure3(4s) 5–8 s (N
Therefore,
7 and Table )
5, the this
3(5s) after the end
paper further
sensitivity of theof
distraction
selected
driver’s as segments
four
control control
operation samples 2–5for
ofreturned comparison.
s (N to ), 3–6
the
3(2s) s As
normal shown
(Nstate
3(3s)), 4–7 ins Figure
after (N3(4s)
the ) 7and
and5–8
distraction Table 5, 3(5s)
s (N
behavior the)ended
sensitivity
after the 5of
for ends. the
of
driver’s control
distraction operation
as control returned
samples to the normal
for comparison. Asstate
shown after
in the distraction
Figure behavior
7 and Table 5, theended for 5 s.
sensitivity of the
driver’s control operation returned to the normal state after the distraction behavior ended for 5 s.
Int. J. Environ. Res. Public Health 2019, 16, 1464 9 of 13
Int. J. Environ. Res. Public Health 2019, 16, x 9 of 13

Eventually,the
Eventually, theoperation
operationofofviewing
viewingresulted
resultedininaadeterioration
deteriorationofofthe
thelongitudinal
longitudinaldriving
driving
controlstate
control (T(34)==−3.265,
state(T(34) −3.265,pp== 0.002; T(34)== 5.709,
0.002; T(34) 5.709, pp << 0.01), while
while the
the lateral
lateral control
controlstate
statewas
wasnot
not
significantlyaffected
significantly affected(F(1.9, 63.2)==1.429,
(F(1.9,63.2) 1.429,pp== 0.247).
0.247).

Figure
Figure 7. Statistical
7. Statistical resultsofofcontrol
results control operation
operationsensitivity
sensitivityduring viewing
during information.
viewing information.
Table 5. Results summary of control operation sensitivity during viewing information.
Table 5. Results summary of control operation sensitivity during viewing information.
N1 –N2 N2 –N3 N1 –N3
Sensitivity Index N1–N2 N2–N3 N1–N3
T(34) =T(34)
Sensitivity Index 2.277= 2.277 T(34)
T(34) = 1.259 T(34) = 2.853
= 1.259 T(34) = 2.853
p = 0.029
p = 0.029 p = 0.217
p = 0.217 p = 0.007 p = 0.007
N2 –N3(2S) N2–N3(2S) N2 –N N2–N3(3S) NN 2–N
2 –N3(4S)
3(4S) N 2-–N3(5S)N —N
2
3(3S) 3(5S)
T(34) = −0.164 T(34) = −0.493 T(34) = −1.788 T(34) = −2.279
T(34) = −0.164 p = 0.871 T(34) = −0.493
p = 0.625 T(34) = −1.788 p = 0.029
p = 0.083 T(34) = −2.279
p = 0.871 p = 0.625 p = 0.083 p = 0.029

4. Discussion
4. Discussion
In the process of using mobile phones, answering, dialing and viewing are the most complex
In the process
manual–visual of using operations.
distraction mobile phones, answering,
The intensity of dialing and viewing
the driver’s are the
longitudinal most activity
control complexis
manual–visual distraction operations. The intensity of the driver’s
decreased during answering and dialing, while the lateral control activities remain normal. longitudinal control activity In
iscontrast,
decreased theduring answering
intensity of lateraland dialing,
control whileis the
activity lateral control
decreased activities
in the process ofremain
viewing, normal.
while theIn
contrast, the intensity
longitudinal of lateralaren’t
control activities control activity is decreased
significantly affected. The in the process
results of viewing,
indicate whilehave
that drivers the
longitudinal control activities aren’t significantly
different self-regulatory strategies in different distractions. affected. The results indicate that drivers have
different
Someself-regulatory
studies show strategies in different
that distracted distractions.
drivers often exhibit self-regulatory behaviors, such as
Some studies show that distracted drivers
reducing driving speed, increasing the distance to the lead often exhibit self-regulatory
vehicle, andbehaviors,
making lesssuchlaneas reducing
changes,
driving
which are considered to reduce the driving demand and accident risk [37-39]. Some studieswhich
speed, increasing the distance to the lead vehicle, and making less lane changes, have are
also
considered to reduce the driving demand and accident risk [37–39]. Some studies
found less lane deviation during hand-held calls and increased lane deviation during texting [40-43], have also found less
lane
whichdeviation during
is consistent hand-held
with calls
the results of and
this increased
paper. Also, lane deviation
some studiesduring
report texting [40–43],
dissimilar resultswhich is
[44,45],
consistent
especiallywith the results
in natural drivingof this paper.
studies Also, This
[33,34]. somemay studies report to
be related dissimilar results [44,45],
the complexity of natural especially
driving
in natural driving studies [33,34]. This may be related to the complexity of
samples. At the beginning of this study, some interfering factors are eliminated by setting sample natural driving samples.
At the beginning
screening of this
conditions and study,
indexsome interferingThe
construction. factors are eliminated
intensity by setting
of the control sample
activity can screening
reflect the
conditions and index
decision-making construction.
willingness The intensity
of drivers. of the control
The results show that activity can are
drivers reflect the decision-making
unable to devote more
willingness of drivers.
attention resources Theface
in the results show
of extra that drivers
workload causedare by
unable
complexto devote more attention
manual–visual resources
distractions, and
intend to reduce longitudinal control activities, so as to ensure that the intensity of lateral reduce
the face of extra workload caused by complex manual–visual distractions, and tend to control
longitudinal
activity is notcontrol activities,
affected. As thesodistraction
as to ensure that the intensity
intensifies, of lateral
the strategy control
shifts. activity
Although thisis behavior
not affected.
is a
As the distraction intensifies, the strategy shifts. Although this behavior is
kind of self-protection of the driver, the longitudinal self-regulatory behaviors of drivers have aa kind of self-protection
significant negative impact on the surrounding traffic flow [46]. Under the condition of high-speed
driving, the influence of this regulation on driving risk in the region needs to be further explored.
Int. J. Environ. Res. Public Health 2019, 16, 1464 10 of 13

of the driver, the longitudinal self-regulatory behaviors of drivers have a significant negative impact
on the surrounding traffic flow [46]. Under the condition of high-speed driving, the influence of this
regulation on driving risk in the region needs to be further explored.
In addition, during the process of answering and dialing, the sensitivity of the driver to perform
control operations is significantly affected, which proves that the distracted operation leads to a duller
response of the driver. From the perspective of the stability of the control state, under the influence of
the above two aspects, the stability of the driver’s longitudinal control state during the answering and
dialing process is deteriorated, but the stability of the lateral control state can still be maintained in a
normal state.
Hang up operation has little influence on the driver’s control behavior. The analysis results show
that although the operation weakens the sensitivity of the driver’s control operation, it doesn’t have
a significant impact on the stability of vehicle control, indicating that the driver can cope with the
distraction caused by this operation.
The analysis results of this paper show that the talking and listening operation has no significant
impact on the intensity of the driver’s control activities, indicating that the distraction exerts less
workload on the driver. However, the effect of cognitive distraction caused by talking and listening on
the driver’s control sensitivity is similar to that of dialing and answering. Moreover, the influence of
talking and listening operation on the stability of the longitudinal control state of the driver is more
severe, indicating that the cognitive distraction will also have a significant impact on the stability of
driving. Compared with other operation processes, the duration distribution of talking and listening
process is longer and more discrete, while the distraction duration may affect the vehicle control
stability to some extent. Therefore, 81 talking and listening operation samples satisfying the screening
conditions (average duration = 50.1 s, variance = 46.1) are extracted to further explore the correlation
between duration and the stability of longitudinal control state. As shown in Table 6, no significant
correlation has been found within the duration range of the current study.

Table 6. Results of Pearson correlation test.

Duration Control State

Duration Pearson Correlation 1 −0.034
p-value 0.760
Control state Number of samples 81 81

In the process of viewing information, viewing also leads to a significant decrease in the sensitivity
of the driver’s integrated control operation. Moreover, different from other operations, the effect of
cognitive distraction caused by the driver’s brain processing information still exists 4–5 s after viewing
operation. According to the analysis results of the control state, although the driver’s lateral control
activities are significantly affected, the stability of the lateral control state is not affected, while the
longitudinal control state shows significant fluctuations. Therefore, this paper argues that maintaining
longitudinal stability requires more effort from the driver than maintaining lateral stability, which
makes the stability of longitudinal control more susceptible to influence.

5. Conclusions
Based on the Shanghai natural driving project, this paper collected high-precision sample data of
distracted driving with mobile phones. Compared with driving simulation experiments and real car
experiments, natural driving study provides data with higher quality and authenticity. In the sample
acquisition, several screening conditions are set to reduce the influence of confounding factors, so as to
increase the reliability of the analysis results. At the same time, based on the theory of ergonomics, this
paper argues that the use of mobile phones mainly impacts the intensity of the driver control activity,
the sensitivity of control operation and the stability of vehicle control state. Furthermore, based on the
concept of moving time window, we make full use of the natural driving sequence data to construct
Int. J. Environ. Res. Public Health 2019, 16, 1464 11 of 13

the characteristic index. Finally, according to the behavior characteristics of different stages in the
process of using mobile phone, this paper classifies the distraction process into different sub-operations
and analyses the influence on driver’s control behavior. The conclusions are as follows.
The study finds that drivers used different resource allocation strategies to deal with the
extra workload caused by distraction under different distraction operations. Typical manual–visual
distraction operations, such as answering, dialing and viewing, can result in a significant interference
in the driver’s attention resources. Therefore, the driver has adopted some coping strategies under
different distraction intensity. The driver tends to decrease the longitudinal control activities in the
process of answering or dialing, and decrease the lateral control activities in the process of viewing,
which plays an important role in reducing driving demand.
In the process of hand-held calls, talking and listening leads to a decrease in the sensitivity of
the driver’s control operation, and the degree of influence is no different from other manual–visual
operations. In addition, talking and listening leads to the most significant reduction in the stability
of longitudinal control state in the process of answering calls. The influence of talking and listening
operation on driver’s control behavior is no less than that of other operations, indicating that banning
only the handheld communication mode cannot eliminate the security risks caused by the cognitive
distraction. The use of hands-free calls should be restricted. Natural driving research for hands-free
calling will be conducted in subsequent studies.
Nowadays, with the development of the functions of smart phones, the behavior of using mobile
phones to edit short messages is less frequent, while the behaviors of browsing and viewing is more
frequent. The advent of social media, mobile navigation and especially ride-hailing services has left
drivers completely dependent on their phones. However, with the proliferation of this behavior,
there is a lack of effective regulation in China. It can be seen that the viewing behavior not only has
a significant influence on the sensitivity and stability of driving control, but also that the influence
will still exist for a period after the distraction behavior ends. Therefore, at least on highways or
expressways, it is necessary to prohibit the use of smartphones to view all kinds of information.
In addition, the research background of this paper is based on the environment of highways
or expressways. Natural driving research on urban roads is more complicated, and needs to be
further studied.

Author Contributions: L.Z. and B.C. conceived and designed the study; B.C. collected and analyzed the data;
L.Z., F.G., and J.W. provided guidance, comments and editorial revisions. L.Z., B.C., M.Y., F.G., and J.W. wrote
the paper.
Funding: This research project was jointly sponsored by Fundamental Research Funds for the Central Universities
(No. 22120180237).
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to
publish the results.

References
1. Violanti, J.M.; Marshall, J.R. Cellular phones and traffic accidents: An epidemiological approach. Accid. Anal.
Prev. 1996, 28, 265–270. [CrossRef]
2. Redelmeier, D.A.; Tibshirani, R.J. Association between cellular-telephone calls and motor vehicle collisions.
N. Engl. J. Med. 1997, 336, 453–458. [CrossRef]
3. Strayer, D.L.; Crouch, D.J.; Drews, F.A. A Comparison of the Cell Phone Driver and the Drunk Driver. J. Hum.
Factors Ergon. Soc. 2006, 48, 381–391. [CrossRef] [PubMed]
4. Rothman, L.; Pike, I.; Belton, K.; Olsen, L.; Fuselli, P.; Macpherson, A. Barriers and Enablers to Enacting Child
and Youth Related Injury Prevention Legislation in Canada. Int. J. Environ. Res. Public Health 2016, 13, 656.
[CrossRef] [PubMed]
5. Dingus, T.A.; Guo, F.; Lee, S.; Antin, J.F.; Perez, M.; Buchanan-King, M.; Hankey, J. Driver crash risk factors
and prevalence evaluation using naturalistic driving data. Proc. Natl. Acad. Sci. USA 2016, 113, 2636–2641.
[CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2019, 16, 1464 12 of 13

6. Zhu, M.; Rudisill, T.M.; Rauscher, K.J.; Davidov, D.M.; Feng, J. Risk Perceptions of Cellphone Use While
Driving: Results from a Delphi Survey. Int. J. Environ. Res. Public Health 2018, 15, 1074. [CrossRef] [PubMed]
7. Guo, F.; Klauer, S.G.; Fang, Y.; Hankey, J.M.; Antin, J.F.; Perez, M A.; Lee, S.E.; Dingus, T.A. The effects of age
on crash risk associated with driver distraction. Int. J. Epidemiol. 2017, 46, 258–265. [CrossRef]
8. Klauer, S.G.; Guo, F.; Simons-Morton, B.G.; Ouimet, M.C.; Lee, S.E.; Dingus, T.A. Distracted Driving and
Risk of Road Crashes among Novice and Experienced Drivers. N. Engl. J. Med. 2014, 370, 54–59. [CrossRef]
9. Ibrahim, J.K.; Anderson, E.D.; Burris, S.C.; Wagenaar, A.C. State laws restricting driver use of mobile
communications devices distracted-driving provisions, 1992–2010. Am. J. Prev. Med. 2011, 40, 659–665.
[CrossRef]
10. Dozza, M.; Flannagan, C.A.; Sayer, J.R. Real-world effects of using a phone while driving on lateral and
longitudinal control of vehicles. J. Safety Res. 2015, 55, 81–87. [CrossRef]
11. Nelson, E.; Atchley, P.; Little, T.D. The effects of perception of risk and importance of answering and initiating
a cellular phone call while driving. Accid. Anal. Prev. 2009, 41, 438–444. [CrossRef]
12. Li, W.; Gkritza, K.; Albrecht, C. The culture of distracted driving: Evidence from a public opinion survey in
Iowa. Transp. Res. Part F Traffic Psychol. Behav. 2014, 26, 337–347. [CrossRef]
13. Atchley, P.; Atwood, S.; Boulton, A. The choice to text and drive in younger drivers: Behavior may shape
attitude. Accid. Anal. Prev. 2011, 43, 134–142. [CrossRef]
14. Metz, B.; Landau, A.; Just, M. Frequency of secondary tasks in driving—Results from naturalistic driving
data. Saf. Sci. 2014, 68, 195–203. [CrossRef]
15. Stutts, J.; Feaganes, J.; Reinfurt, D.; Rodgman, E.; Hamlett, C.; Gish, K.; Staplin, L. Driver’s exposure to
distractions in their natural driving environment. Accid. Anal. Prev. 2005, 37, 1093–1101. [CrossRef]
16. Fitch, G.M.; Bartholomew, P.R.; Hanowski, R.J.; Perez, M.A. Drivers’ visual behavior when using handheld
and hands-free cell phones. J. Safety Res. 2015, 54, 105–108. [CrossRef]
17. Strickland, D.L. Visual-Manual NHTSA Driver Distraction Guidelines for in-Vehicle Electronic Devices; National
Highway Traffic Safety Administration: Washington, DC, USA, 2012.
18. Bao, S.; Guo, Z.; Flannagan, C.; Sullivan, J.; Sayer, J.R.; LeBlanc, D. Distracted Driving Performance Measures.
Transp. Res. Rec. 2015, 2518, 68–72. [CrossRef]
19. Hagiwara, T.; Sakakima, R.; Hashimoto, T.; Kawai, T. Effect of distraction on driving performance using
touch screen while driving on test track. Intell. Veh. Symp. 2013, 1149–1154. [CrossRef]
20. Wolfe, J.M.; Horowitz, T.S.; Kenner, N.M. Cognitive psychology: Rare items often missed in visual searches.
Nature 2005, 435, 439–440. [CrossRef]
21. Wickens, C.D. Multiple resources and mental workload. Hum. Factors 2008, 50, 449–455. [CrossRef]
22. Collet, C.; Clarion, A.; Morel, M.; Chapon, A.; Petit, C. Physiological and behavioral changes associated to
the management of secondary tasks while driving. Appl. Ergon. 2009, 40, 1041–1046. [CrossRef]
23. Haigney, D.E.; Taylor, R.G.; Westerman, S.J. Concurrent mobile (cellular) phone use and driving performance:
Task demand characteristics and compensatory processes. Transp. Res. Part F Traffic Psychol. Behav. 2000, 3,
113–121. [CrossRef]
24. Caird, J.K.; Johnston, K.A.; Willness, C.R.; Asbridge, M. The use of meta-analysis or research synthesis to
combine driving simulation or naturalistic study results on driver distraction. J. Safety Res. 2014, 49, 91–96.
[CrossRef]
25. Rakauskas, M.E.; Gugerty, L.J.; Ward, N.J. Effects of naturalistic cell phone conversations on driving
performance. J. Safety Res. 2004, 35, 453–464. [CrossRef]
26. Beede, K.E.; Kass, S.J. Engrossed in conversation: The impact of cell phones on simulated driving performance.
Accid. Anal. Prev. 2006, 38, 415–421. [CrossRef]
27. Hosking, S.G.; Young, K.L.; Regan, M.A. The effects of text messaging on young drivers. Hum. Factors 2009,
51, 582–592. [CrossRef]
28. Strayer, D.L.; Drews, F.A.; Johnston, W.A. Cell phone-induced failures of visual attention during simulated
driving. J. Exp. Psychol. Appl. 2003, 9, 23–32. [CrossRef]
29. Strayer, D.L.; Johnston, W.A. Driven to distraction: Dual-Task studies of simulated driving and conversing
on a cellular telephone. Psychol. Sci. 2001, 12, 462–466. [CrossRef]
30. Drews, F.A.; Yazdani, H.; Godfrey, C.N.; Cooper, J.M.; Strayer, D.L. Text messaging during simulated driving.
Hum. Factors 2009, 51, 762–770. [CrossRef]
Int. J. Environ. Res. Public Health 2019, 16, 1464 13 of 13

31. Young, R. Talking on a Wireless Cellular Device While Driving: Improving the Validity of Crash Odds Ratio
Estimates in the SHRP 2 Naturalistic Driving Study. Safety 2017, 3, 28. [CrossRef]
32. Xiong, H.; Bao, S.; Sayer, J.; Kato, K. Examination of drivers’ cell phone use behavior at intersections by using
naturalistic driving data. J. Safety Res. 2015, 54, 89–93. [CrossRef]
33. Tivesten, E.; Dozza, M. Driving context influences drivers’ decision to engage in visual-manual phone tasks:
Evidence from a naturalistic driving study. J. Safety Res. 2015, 53, 87–96. [CrossRef]
34. Schneidereit, T.; Petzoldt, T.; Keinath, A.; Krems, J.F. Using SHRP 2 naturalistic driving data to assess drivers’
speed choice while being engaged in different secondary tasks. J. Safety Res. 2017, 62, 33–42. [CrossRef]
35. Guo, F. Statistical methods for naturalistic driving studies. Ann. Rev. Stat. Appl. 2019, 6, 309–328. [CrossRef]
36. Ye, M.; Osman, O.A.; Ishak, S.; Hashemi, B. Detection of driver engagement in secondary tasks from observed
naturalistic driving behavior. Accid. Anal. Prev. 2017, 106, 385–391. [CrossRef]
37. Oviedo-Trespalacios, O.; Haque, M.M.; King, M.; Demmel, S. Driving behaviour while self-regulating mobile
phone interactions: A human-machine system approach. Accid. Anal. Prev. 2018, 118, 253–262. [CrossRef]
38. Young, K.L.; Osborne, R.; Koppel, S.; Charlton, J.L.; Grzebieta, R.; Williamson, A.; Senserrick, T. What
are Australian drivers doing behind the wheel? An overview of secondary task data from the Australian
Naturalistic Driving Study. J. Australas. Coll. Road Saf. 2019, 30, 27–33.
39. Oviedo-Trespalacios, O.; Haque, M.M.; King, M.; Washington, S. Self-regulation of driving speed among
distracted drivers: An application of driver behavioral adaptation theory. Traffic Inj. Prev. 2017, 18, 599–605.
[CrossRef]
40. Garrison, T.M.; Williams, C.C. Impact of Relevance and Distraction on Driving Performance and Visual
Attention in a Simulated Driving Environment. Appl. Cong. Psychol. 2013, 27, 396–405. [CrossRef]
41. Reimer, B.; Mehler, B.; Donmez, B. A study of young adults examining phone dialing while driving using a
touchscreen vs. a button style flip-phone. Transp. Res. Part F Traffic Psychol. Behav. 2014, 23, 57–68. [CrossRef]
42. McKeever, J.D.; Schultheis, M.T.; Padmanaban, V.; Blasco, A. Driver performance while texting: Even a little
is too much. Traffic Inj. Prev. 2013, 14, 132–137. [CrossRef]
43. Rudin-Brown, C.M.; Young, K.L.; Patten, C.; Lenne, M.G.; Ceci, R. Driver distraction in an unusual
environment: Effects of text-messaging in tunnels. Accid. Anal. Prev. 2013, 50, 122–129. [CrossRef]
44. Irwin, C.; Monement, S.; Desbrow, B. The influence of drinking, texting, and eating on simulated driving
performance. Traffic Inj. Prev. 2015, 16, 116–123. [CrossRef]
45. Cao, S.; Liu, Y. Concurrent processing of vehicle lane keeping and speech comprehension tasks.
Accid. Anal. Prev. 2013, 59, 46–54. [CrossRef]
46. Stavrinos, D.; Jones, J.L.; Garner, A.A.; Griffin, R.; Franklin, C.A.; Ball, D.; Welburn, S.C.; Ball, K.K.;
Sisiopiku, V.P.; Fine, P.R. Impact of distracted driving on safety and traffic flow. Accid. Anal. Prev. 2013, 61,
63–70. [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/).