Transportation Research Part C 77 (2017) 351–365

Contents lists available at ScienceDirect

Transportation Research Part C

journal homepage: www.elsevier.com/locate/trc

Modelling driver distraction effects due to mobile phone use on

reaction time
Pushpa Choudhary, Nagendra R. Velaga ⇑
Transportation Systems Engineering, Civil Engineering Department, Indian Institute of Technology (IIT) Bombay, Mumbai 400076, India

a r t i c l e i n f o a b s t r a c t

Article history: Phone use during driving causes decrease in situation awareness and delays response to
Received 2 March 2016 the events happening in driving environment which may lead to accidents. Reaction time
Received in revised form 24 January 2017 is one of the most suitable parameters to measure the effect of distraction on event detec-
Accepted 7 February 2017
tion performance. Therefore, this paper reports the results of a simulator study which anal-
Available online 16 February 2017
ysed and modelled the effects of mobile phone distraction upon reaction time of the Indian
drivers belonging to three different age groups. Two different types of hazardous events:
Keywords:
(1) pedestrian crossing event and (2) road crossing event by parked vehicles were included
Driver distraction
Reaction time
for measuring drivers’ reaction times. Four types of mobile phone distraction tasks: simple
Conversation conversation, complex conversation, simple texting and complex texting were included in
Texting the experiment. Two Weibull AFT (Accelerated Failure Time) models were developed for
Survival analysis the reaction times against both the events separately, by taking all the phone use condi-
tions and various other factors (such as age, gender, and phone use habits during driving)
as explanatory variables. The developed models showed that in case of pedestrian crossing
event, the phone use tasks: simple conversation, complex conversation, simple texting and
complex texting caused 40%, 95%, 137% and 204% increment in the reaction times and in
case of road crossing event by parked vehicles, the tasks caused 48%, 65%, 121% and
171% increment in reaction times respectively. Thus all the phone use conditions proved
to be the most significant factors in degrading the driving performance.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction

The World Health Organization (WHO) recognizes road accidents as the 9th leading cause of deaths, which accounts for
2.2% of the total deaths worldwide (WHO fact sheets, 2014). In India, the road accident statistics reveal that there is an acci-
dent happening every minute and causing a loss of life in every four minutes (MRTH, 2013). There are evidences that indicate
driver distraction (caused by digital communication devices such as Mobile phones or in-vehicle devices such as GPS) as one
of the main reasons of accidents; for instance McEvoy and Stevenson (2007) conducted a case crossover study in two years of
time span in Perth, Western Australia and found that driver distraction due to phone use, contributed to 14% of the road
crashes. Similarly it was reported that 10%, 37%, 10.7% and 16% accidents in New Zeeland, Spain, Canada and US were caused
by driver distractions (NRSC, 2010; DGT, 2008; World Health Organization, 2011 and Ascone et al., 2009). In India one online
survey conducted on risk assessment of mobile phone use by Shabeer and Banu (2012) revealed that 31% of the drivers who
used a mobile phone during driving met with accidents. Further, analysis on sources of these distractions showed that use of

⇑ Corresponding author at: Civil Engineering Department, IIT Bombay, India.

E-mail address: n.r.velaga@iitb.ac.in (N.R. Velaga).

http://dx.doi.org/10.1016/j.trc.2017.02.007
0968-090X/Ó 2017 Elsevier Ltd. All rights reserved.
352 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

digital devices such as: mobile phones, GPS and handling of entertainment devices are the commonly prevailing sources of
distraction; however, among all the distractions, mobile phone use is the most common one while driving (Horberry et al.,
2006). Despite the ban on mobile phone use during driving, the proportion of drivers with mobile phone use is reasonably
high; for example, in some metro cities in Australia, US, UK, Spain, Canada and Sweden it is 4.75%, 5.8%, 3.7%, 1.3%, 5.9% and
30% respectively (Young et al., 2010; Eby et al., 2006; Sullman et al., 2015; Prat et al., 2015; Burns et al., 2008; Thulin and
Gustafsson, 2004). The use of mobile phone is not only limited to talking, but also extended to sending (or reading) messages
and emails. In fact, it has been reported that mobile phone usage for texting while driving is as high as, 45%, 16.67%, and 27%
in the United Kingdom, Australia and US respectively (World Health Organization, 2011). This trend is more common in
young drivers; for example, a study in Australia showed that 58% of young drivers read text messages and 37% of them send
the messages during driving (World Health Organization, 2011).
Driving performance, particularly reaction time (response to the events happening in driving environment), is adversely
affected by mobile phone usage during driving and results in increased risk of accident involvement. Redelmeier and
Tibshirani (1997) showed that if a driver is talking on a phone during driving then accident occurring probability increases
four times compared to without phone use driving condition. McKnight and McKnight (1993) speculated that demands on
the driver’s attention and skill gets increased due to mobile phone use, which results in increased reaction time and
increased failure rates to detect and react according to the traffic signals and hazardous events happening on the road.
As summarized above, distraction has become major road safety issue in the last two decades and many studies have
been conducted in this direction world-wide. Present study aims to investigate the distraction effects due to mobile phone
use for Indian driving conditions, where the driving conditions and driver behaviours are completely different from others.
The study examines both, cognitive and visual distraction effects (with different level of complexities) on driver reaction
time.

2. Literature review

Lee et al. (2008) defined driver distraction as diversion of attention away from activities critical for safe driving towards a
competing activity. Existing literature has shown that both cognitive and visual distraction can impair the driver’s reaction
time behaviour (Bellinger et al., 2009). Following subsections summarize the previous study results of mobile phone use on
reaction time of a driver, followed by different experimental setups and analysis methods adopted by previous studies.

2.1. Effects of phone use on reaction time

Many studies illustrated about the increase in reaction time due to phone usage (Al-Darrab et al., 2009; Alm and Nilsson,
1994; Brookhuis et al., 1991; Strayer and Drew, 2004). Literature has showed that both conversation and texting impair the
event perceiving abilities of drivers, which leads to unsafe driving conditions (Bellinger et al., 2009; Redelmeier and
Tibshirani, 1997; Lamble et Al., 1999). meta-analysis studies conducted on effects of mobile phone use during driving,
strongly emphasized on the fact that reaction time significantly increases if a driver is concurrently involved in conversation
(Horrey and Wickens, 2006) or texting (Caird et al., 2014).
Caird et al. (2008) critically reviewed 33 studies on effects of phone use and concluded that reaction time increases by
0.25 s while talking on phone during driving. Lee et al. (2001) analysed the effects of cognitive loading of a speech based
e-mail system on driving performance and found that there is 30% increment in reaction-time. Cooper et al. (2011) observed
that reaction time increases to two folds if a phone is used for texting while driving. Klauer et al. (2006) and Yannis et al.
(2014) observed increment in accident probabilities when texting on phone during driving.
Various experimental designs have been used for estimating the reaction time, for example, Haque and Washington
(2014) analysed driver’s reaction time against a simulated event, where a pedestrian entered a zebra crossing from a side-
walk. Results of the study exhibited 40% increment in reaction time when the drivers were talking on the phone during driv-
ing. Leung et al. (2012) examined the reaction time against a hazardous event of sudden appearance of a truck in front of the
subject vehicle and found significant increment in reaction time while conversing on phone. Consiglio et al. (2003) measured
the reaction time for critical traffic signals and reported that 72 ms increment happened in reaction time of the drivers when
they were engaged in phone conversation. Patten et al. (2004) used Peripheral Detection Task (PDT) for examining driver’s
reaction time and found that detection of peripheral stimuli got delayed 72 ms and 261 ms when the driver was involved in
simple and complex conversation on phone during driving respectively. Recarte and Nunes (2003) measured drivers’
response by a simultaneous visual-detection and discrimination test and observed significant reduction in event detection
performance.
Driver’s demographic characteristics (age and gender) have also been considered while analysing the reaction time in dis-
tracted driving conditions. Haque and Washington (2014) observed that one year increment in driver’s age caused 12% incre-
ment in driver’s reaction time. In a meta-analysis study by Caird et al. (2008) it was concluded that older drivers were more
affected (reaction time increment was 0.46 s) when compared to younger drivers (reaction time increment was 0.19 s) while
conversing on a phone during driving. Similarly, in a field study, Hancock et al. (2003) showed that the impairment in reac-
tion time was 0.18 s higher for older drivers than the younger drivers. This study also showed that the reaction time
increased by 0.25 s for female drivers while for male drivers the increment observed was 0.14 s. Several studies have also
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 353

accounted for different modes of phones (hands-free and handheld) while analysing the distraction effects on reaction time
and found that both the user interfaces had equal adverse effect on reaction time of the drivers (Matthews et al., 2003;
Horrey and Wickens, 2006). In case of different road environments, such as: rural, urban or motorway, reaction time was
significantly affected by the complexity levels of the environment (Benedetto et al., 2012; Törnros and Bolling, 2006).

2.2. State of the art research on experimental setup and analysis methods

Two basic approaches for conducting the distraction studies for reaction time measurement are: (1) Field testing and (2)
Simulator testing (Hancock et al., 2003; Törnros and Bolling, 2006; Jamson and Merat, 2005). Though some of the researchers
(Harms and Patten, 2003; Hancock et al., 2003; Patten et al., 2004) adopted field study approach, but, safety is the main issue
in conducting the distraction studies in field. Therefore, a large number of researchers used driving simulators for their study
(Jamson and Merat, 2005; Törnros and Bolling, 2006; Beede and Kass, 2006; Peng et al., 2014). Because, it is safe to collect
data in laboratory (in controlled environment) than in field; and moreover, data collection and extraction become more pre-
cise and easy for simulator approach.
For analysing the data on driving performance, different methods have been used by researchers. Existing studies used T-
test (Benedetto et al., 2012), Wilcoxon test (Leung et al., 2012), Contingency table method (Huth et al., 2015) and ANOVA
(analysis of variance) (Patten et al., 2004; Santos et al., 2005) to check whether the distraction is causing any significant effect
on driving performance. Very few of them used statistical modelling such as: Linear mixed model (Peng et al., 2014) and
parametric duration models (Haque and Washington, 2015) for actually finding out the relationship between all explanatory
variables and driving performance.
Despite the vast availability of literature on distraction effect of mobile use on driving performance, very few studies com-
pared the effects of different types of phone use (cognitive and visual) and different levels of complexity involved in phone
use (Leung et al., 2012; Patten et al., 2004). Caird et al. (2008) showed that the increment in reaction time was higher for
cognitively demanding conversation (i.e., 0.33 s) when compared to naturalistic conversation (i.e., 0.14 s). Leung et al.
(2012) showed that texting had worse effect on reaction time compared to conversation task. Table 1 summarizes the liter-
ature review of distraction studies on visual and cognitive distraction effects due to phone use.

2.3. Research motivation

As seen from the above mentioned literature review, most of the studies on mobile phone distraction are conducted in US,
Canada and other developed countries; but despite the huge prevalence of mobile phone use, very few distraction studies are
conducted in developing countries. The present study attempts to examine the distraction effect on reaction time in Indian
driving conditions particularly in rural road environment. Moreover, the present study quantitatively differentiated on the
effects of both the conversation and texting tasks on driver’s reaction time, which is the additive contribution to the existing
literature where most of the studies either concentrated on conversation effects or on the texting effects. Additionally, dif-
ferent levels of complexity of the conversation and texting tasks are introduced in the present study, which reflect the true
driving behaviour in presence of low or high level of distraction. For analysis purpose, instead of the traditional approach of
performing ANOVA test and linear regression tests, a statistical modelling approach with the survival analysis technique
(Weibull AFT) is adopted for modelling the reaction time of the driver, which is also suggested by Haque and Washington
(2014).

3. Methodology

Fig. 1 shows the methodology adopted for conducting the present study. The steps adopted to perform the present study
are as following (for more details please see Choudhary and Velaga, 2017):

(1) Scenario development: Firstly, simulated road scenarios which consist of simple road geometry and traffic conditions,
are designed in the simulator with the help of SimCreator and SimVista software. And then two real -world events are
incorporated in the scenarios.
(2) Secondary task design: The secondary tasks are designed for creating both the types of distractions due to phone use
during driving: (1) visual (texting) and (2) cognitive (conversation). Both of them are split into two different levels of
complexities.
(3) Participant identification and questionnaire preparations: In total, hundred licenced drivers are identified for data collec-
tion. A questionnaire is prepared for collecting information on driver demographic details, driving characteristics and
phone use habits, etc.
(4) Data collection: The data for each participant is collected for simulated drive in two conditions: non-distracted (driving
without any phone use) and distracted drive (driving with concurrent phone use).
354 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

Table 1
State-of-the art literature on distraction effects (increased reaction time) due to mobile phone use.

Study Phone task Event Participants Test scenario/field route Analysis Findings
technique
Alm and Nilsson Conversation Sudden 40 Two routes, each of 80 km ANOVA Reaction time got increased for
(1994) appearance length easy route but did not change
of visual significantly for hard route
stimuli on (1) simple road geometry
wind screen (2) complex geometry
with more horizontal
and vertical curves
Alm and Nilsson Conversation 8 critical 40 80 km route, straight and ANOVA Increase in reaction time from
(1995) events easy to drive 0.38 s to 0.56 s for young drivers
performed from 0.43 s to 1.46 s for the
by the lead older drivers
vehicle
Consiglio et al. Conversation Response to 22 Not based on actual driving ANOVA Increased mean reaction time
(2003) activation task
of red brake
lamp
Patten et al. (2004) Simple and Peripheral 40 74 km long Motorway ANOVA Reaction time increased in both
Complex detection (Field study) conversation types; and
Conversations task difficulty level of phone task
found significant
Beede and Kass Conversation Signal 36 8.69 km long route ANOVA Response to stop sign and traffic
(2006) detection including intersections, light got delayed in conversation
task buildings, pedestrians, cars, task
and other obstacles
Törnros and Conversation Peripheral 48 Two rural road Effect size Reaction time increased in all
Bolling (2006) detection environments with estimation driving environments in almost
task different speed limits, (according to similar manner
Three urban road Omega2)
environments with
different complexity level
Bellinger et al. Conversation Response to 27 Not based on actual driving ANOVA Increased mean reaction time
(2009) activation task
of red brake
lamp
Leung et al. (2012) Conversation Sudden 12 20 min route having Wilcoxon- Braking reaction time was
Texting appearance straight road and curves signed rank highest for texting during
of truck in (night time scenario) test driving followed by conversing
front of during driving and least in no
subject phone use driving conditions
vehicle
Yannis et al. Texting Animal 34 Urban and rural road Lognormal Reaction time increased
(2014) crossing in environments regression significantly in both urban and
front of rural environment
subject
driver
Haque and Conversation A 32 7 km long and included a Survival Reaction time increased in the
Washington pedestrian detailed simulation of the analysis presence of conversation task
(2014) entering a Brisbane CBD and a (Weibull AFT
zebra hypothetical suburban area model with
crossing created gamma
from heterogeneity)
sidewalk

(5) Analysis and Results: For estimating the distraction effects on driving performance, firstly preliminary analysis of the
collected data is done, and then statistical modelling of reaction times is performed with the help of a parametric sur-
vival analysis technique. Results of the developed models are interpreted in terms of the significant effects of any fac-
tor (explanatory variables) on reaction time.

Each step is further explained in details in the following sections.

4. Data collection

Driving performance of the participants was measured through various simulator experiments. The data on performance
were collected in terms of reaction time, speed, lateral performance, etc. (further details on data collection are available in
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 355

Literature Review

Scenario design Secondary task design Analysis

1. Pedestrian crossing event Visual Distraction Techniques
2. Road crossing event by parked vehicle 1. Simple texting
2. Complex texting
Cognitive Distraction
Scenario Development 1. Simple conversation
(In Driving Simulator) 2. Complex conversation

Participant identification Data collection

1. Distracted driving
Questionnaire preparations 2. Non-distracted driving

Preliminary Analysis of data

Statistical Modelling with

parametric survival model

Results

Fig. 1. Study methodology for finding distraction effects of mobile use on driver reaction time.

Choudhary and Velaga, 2017). The present study focused on reaction time behaviour of the drivers; therefore details of the
data collection process for the same are described in the following subsections.

4.1. Driving simulator

The driving simulator used for the study is a fully instrumented open cab driving simulator, providing 150° horizontal
view on three 42 in. LED systems (Fig. 2). The cab includes power steering (with active force feedback), brake, gear selector
(for manual transmission), turn signals and flexible button inputs for experimental usage, thus exhibited in appropriate vehi-
cle behaviour in the simulator.

4.2. Development of test scenarios and events

The simulated route contained an undivided two-way road with two lanes in each direction of travel, which is similar to
rural national highways in India. The participants drove for five driving conditions: No phone use, Simple conversation,

Fig. 2. Driving simulator used for the study.

356 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

Complex conversation, Simple texting and Complex texting. For each condition, drive was 3.5 km long and had one haz-
ardous event to measure human reaction time. Instead of giving five repetitive drives on the same route, the four drives
(except no phone use drive), were combined in a single closed loop road network. Two different types of unexpected haz-
ardous events were included: (1) a group of pedestrians crossing the road (crossing at random place i.e. not at zebra crossing)
and (2) road crossing by parked vehicles from road side. In the pedestrian crossing event, pedestrians started crossing the
road in front of the subject vehicle when it was 130 m away from them and in the road crossing event, parked vehicles
(a car and a truck) started crossing the road in the same manner as stated above in order to join the oncoming traffic stream.
Reaction time was measured as the time difference between the time when event started and the time when the driver first
responded to the event by releasing the accelerator (if the driver was accelerating just before the detection of the event) or
pressing the brake pedal (if the driver was not pressing the accelerator pedal just before the detection of the event). For most
of the drivers, the release of accelerator pedal was the first response, but for very few profiles it was observed that the drivers
were not pressing the accelerator pedal at the time of start of the event, so pressing the braking pedal was considered as the
first reaction. Fig. 3 demonstrates the events included in the present study.

4.3. Secondary task

Two types of secondary tasks were included in distracted drive: Cognitive task: Hand-held mobile phone conversation
with two difficulty levels (i.e., simple conversation and complex conversation) and Visual task: Text messages with two dif-
ficulty levels (simple texting and complex texting). For simple conversation, simple questions like ‘‘Where did you go for your
last trip?” were asked (Choudhary and Velaga, 2017); while for complex conversation arithmetic problems (multiplication
and addition) and logical puzzle questions were asked (some of them were modified from Burns et al., 2002). For simple tex-
ting tasks the drivers replied in short texts (up to 10 characters) for the questions like ‘‘What is your favourite game?” and
complex texting tasks they replied in longer texts for questions like ‘‘Your address?” (more than 10 characters) (Peng
et al., 2014).

4.4. Participants

Hundred licensed drivers of three different age groups: young (age < 30), mid- age (30
age < 50) and old age (age  50),
participated in the study. Out of these hundred participants, 27 were professional drivers, working in a transport company.
One questionnaire containing three sections: driver’s demographic characteristic, driving characteristic and about mobile
phone use habits (in everyday life as well as during driving) was prepared. Table 2 shows the details of explanatory variables
which were defined based on the information from the questionnaire survey.

4.5. Procedure

Firstly, each of the participants was briefed about the present study. Then the demographic details, driving characteristics
details and details regarding the phones use habits of each of the participants were filled in the questionnaire. After that,
each participant was given a trial drive on the simulator and when the driver got familiar with simulator use, the two drives
used for the actual experiment, were presented to the driver. One of these two drives was completed in driving only condi-
tion (no phone use), while in the second drive four kinds of distraction tasks (simple conversation, complex conversation,
simple texting and complex texting) were presented to the driver, along with the driving task. The order of presenting
the distracted and non-distracted drives was randomised. Additionally the order of the secondary tasks presented in dis-
tracted drive was also randomised.

Fig. 3. Pedestrian crossing event and road crossing event by parked vehicle.
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 357

Table 2
Statistical details obtained from the questionnaire.

Variables Description Type* Levels/values Mean SD Percentage

Driver demographics
Age Driver’s age Cat Young** 24.14 2.79 55
(years)
Mid-age 36.05 5.43 39
(years)
Old-age 54.67 5.04 6
(years)
Gender Driver’s gender Cat Female** 13
Male 87
Driving characteristics
Driving experience How many years of driving experience the driver has Con Young 4.50 3.36
(in years) Mid-age 11.92 8.17
Old-age 34.33 5.89
License owning How long has the driver been in possession of the driving license Con In years 9.078 8.81
period
Trip type Type of trip made generally Cat Work** 40
Recreational 60
Traffic offence due to Whether the driver had been fined for using mobile phone in the Cat No** 100
phone use last three years Yes 0
Accidents Whether the driver met with an accident due to distraction by Cat No** 100
mobile phone use in the last three years Yes 0
Mobile phone habits
Mobile phone use in day to day life
Phone type Which type of mobile phone the driver uses Cat Touch 87
screen**
Normal 13
Calls How many calls the driver receives/makes in a day Con In numbers 13 15.79 –
Text messages How many text messages the driver writes/reads in a day Con In numbers 60 97.97 –
Mobile phone use during driving
(a) Conversation
Frequently Whether the number of calls received = at least three per day Cat No/ Yes 36
conversing
Less frequently Whether the number of calls received = at most two per day Cat No/Yes 25
conversing
Rarely conversing ** Whether the number of calls received = one or two per week Cat No/Yes 39
(b) Texting
Frequently texting Whether the number of texts sent = at least three per day Cat No/Yes 4
Less frequently Whether the number of texts sent = once or twice per week Cat No/Yes 12
texting
Rarely texting** Whether the number of texts sent = once or twice per month Cat No/Yes 84
*
Type: con = continuous variable, cat = categorical variable.
**
Reference categories.

5. Analysis and results

5.1. Preliminary analysis of explanatory variables

Preliminary analysis such as mean, standard deviation (SD) and percentage of the explanatory variables obtained from
the questionnaire data, are presented in Table 2. Here percentage is showing the proportion of driver population for each
category of a categorical variable; for instance, in age variable driver population is distributed as 55% young drivers, 39%
mid-age drivers and 6% old-age drivers. The average driving experience observed was 4.5, 11.9 and 34 years for young,
mid-age and older drivers respectively. It was found that despite the ban on phone use during driving, 36% drivers told that
they respond to all the calls they get while driving (frequently conversing) and 25% drivers respond to one or two calls per day
during driving (less frequently conversing), while only 39% drivers rarely respond any call during driving. This shows that
larger population of drivers are involved in talking on phone.

5.2. Preliminary analysis of dependent variable

Driver’s reaction time depends on many factors such as: Personal characteristics (age, gender, etc.), types of event
(expected or unexpected) and driver’s attention to driving (fully attentive to the driving task or distracted towards any sec-
ondary task during driving). Though in the present study, drivers did not know about exact location of the hazardous events,
but they knew that at any point of their driving they may face some hazardous event, to which they have to react accordingly
as they do in their real-life driving. As mentioned before, driver’s attention to driving was varied in five different levels: fully
358 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

attentive (no distraction), distracted towards simple conversation, complex conversation, simple texting and complex
texting.
To see the changes in driver’s performance due to cognitive and visual distraction, mean reaction times of all the drivers
for distracted driving are plotted along with the non-distracted driving condition (no phone use) in Fig. 4. From this figure, a
vertical shift in reaction times can be seen in both conversation and texting tasks. But, the vertical shifts in reaction times
during the texting tasks are larger compared to the conversation tasks and hence shows that reaction time performance
deteriorates more due to visual distraction.
But, mean reaction times may not be a true representation of resultant value; for instance, AASHTO (American Association
of State Highway and Transportation Officials) assumes reaction time value as 2.5 s, which is comparable to 90th percentile
value of reaction times (not the mean of reaction times) (AASHTO, 2011). Many highway design standards, including Indian
Road Congress (IRC 66) also follows the same standard as set by AASHTO (IRC 66, 1976). The standard value of reaction time
(2.5 s) has been adopted by AASHTO after reviewing some studies on estimation of reaction times. Summaries of these stud-
ies are shown in Table 3.
To account for complex road geometry, all kinds of driver behaviours and complex traffic conditions, a value as 2.5 s was
assumed sufficient by AASHTO, which is more than the 90th percentile of the reaction times of all studies. All these studies
were conducted few decades back, so for identifying the standards for reaction time, driver distraction was not taken into
account because it was not considered as an important factor during that time. But, as on today, driver distraction is very
much prevalent, especially for younger drivers; therefore, the detailed preliminary analysis was done on reaction times
for no phone use and different phone use conditions.
Fig. 5 shows the boxplots for reaction times in all driving conditions. Minimum and maximum reaction times are shown
by lower and upper whisker respectively. 90th and 10th percentiles of reaction times are shown as upper and lower bound-
aries of the box while middle line shows the 50th percentile (median) of reaction time values.
The results show that for the no phone use conditions the results are in line with all previous studies; but, despite the
study being conducted for simple road geometry (simulated) and normal traffic conditions, the 90th percentile of all phone
use conditions except for simple conversation task are higher than 2.5 s.
From this preliminary analysis it can be seen that distracted driving condition increases the reaction time of drivers; fur-
ther, statistical modelling was also done to see that whether only distraction was causing the changes in reaction time or
other factors like age, gender, and phone use habits during driving, also played any crucial role in changing the response
behaviour of the drivers.

5.3. Modelling of reaction times

Statistical modelling was done with survival analysis, which is a set of statistical analysis techniques where the depen-
dent variable is the duration (time) until any particular event (such as occurrence of accident of a vehicle) happens. Basic
terminologies for the survival analysis for present study are defined as following (Washington et al., 2011):

3.00
2.72
Mean Reacon Times (s)

2.50
2.56

2.00 1.84
1.50 1.49

1.00 0.99

0.50
0.00
No Phone Simple Conversaon Complex Simple Texng Complx Texng
Conversaon

Fig. 4. Mean reaction times of drivers in all phone use conditions.

Table 3
Summary of previous studies (referred by AASHTO) on reaction time.

Study Reaction times (summary of results)

Johansson and Rumar (1971) Expected events Unexpected events
Median = 0.6 s 1 s increase from expected event Reaction time
For 10% drivers >1.5 s
Massachusetts Highway accident survey, CWA, ERA report (1935) Average = 0.64 s
For 5% driver >1 s
Normann (1953) Range 0.4–1.7 s
Fambro et al. (1997) (validated the 2.5 s as reaction time value) Average = 1.1 s
90th percentile R.T. = 2.0 s
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 359

Fig. 5. Boxplots of reaction times of all phone use conditions.

Event: Here an event is the detection of the pedestrians or vehicles, which were crossing the road in front of subject
vehicle.
Duration variable is the time elapsed until an event happens. For present study reaction time of driver (i.e., time taken by
the subject driver to detect the event) is the duration variable and the event got ended once the moving pedestrians or vehi-
cles were detected by the subject driver. The duration variable is a continuous random variable T with a cumulative function
F(t) and probability density function f(t). They are formulated as follows (Washington et al., 2011):
Z t
FðtÞ ¼ PðT < tÞ ¼ f ðtÞdt ð1Þ
0

Hazard rate: It gives the conditional probability of time duration ending at time t, given that the duration has continued
until time t (Washington et al., 2011). In this study, as the time increases, the probability of detecting a pedestrian also
increases so hazard function is a monotonous increasing function and can be formulated as following:

f ðtÞ
hðtÞ ¼ ð2Þ
½1  FðtÞ
Survival rate (SðtÞ): It gives the probability of duration variable (reaction time in present study) being greater than some
specified time ‘‘t” i.e. the probability of not detecting the hazardous event (Washington et al., 2011).
SðtÞ ¼ PðT P tÞ ¼ 1  FðtÞ ð3Þ
Present study focused on parametric survival analysis where, effects of covariates are also included in the analysis. There are
two approaches of incorporating covariates in the model (Washington et al., 2011): (1) Proportional hazard based (PH). (2)
Accelerated failure time (AFT). In proportional hazard method, the effects of covariates are multiplicative to the base line
hazard function (where no covariates are included). While for Accelerated failure time model, the effects of covariates rescale
the duration variable directly in base line survival function and produce better results when there is more than one covariate.
For the present study AFT model was taken for the analysis.
For complete parametric solution to the above equations, the type of distribution for duration variable is also needed,
which is assumed as Weibull distribution in the present study; because it can handle monotonic increasing and decreasing
hazard rates (Washington et al., 2011) and hazard rate is monotonous increasing in the present study. So probability density
function of duration variables can be written with two parameters: k location and p scale parameter (Washington et al.,
2011).

f ðtÞ ¼ kpt p1 exp ðktÞp ð4Þ

And corresponding survival rates are given by

SðtÞ ¼ exp ðktÞp ð5Þ

To incorporate Weibull distribution in AFT model the regression equation for T is given by

log T ¼ b0 þ bX þ re ð6Þ
where r is assumed to follow some extreme value function such that T leads to Weibull distribution. The above mentioned
model is valid for independent observations while for the present study, five repeated measures of reaction time were col-
360 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

lected for each driver and thus the reaction time values might be correlated at individual levels. In order to account for this
individual level heterogeneity or frailty, two possible solutions are considered. One way is to model the duration time with
clustered heterogeneity, in this method the model is adjusted for heterogeneity by adjusting the standard error estimates
(Cleves, 2008). The second way is to incorporate shared frailty in the model, which is considered to be the best solution
to handle the unobserved heterogeneity (Haque and Washington, 2014; Gutierrez, 2002). The shared frailty model can be
described as follows (Haque and Washington, 2014; Gutierrez, 2002):

hij ðtjai Þ ¼ /i hij ðtÞ ¼ /i h0 ½t expðbX ij Þ expðbX ij Þ

where hij is the hazard function for ith driver in the jth drive; X ij is a vector of explanatory variables; b is a vector con-
taining coefficient of parameters and ai is the gamma distributed shared frailty with mean one and variance h (Haque
and Washington, 2014).
As the present study had two types of events (pedestrian crossing event and road crossing event by parked vehicles) two
separate models for both the events were developed. The Weibull AFT models with gamma frailty and with clustered hetero-
geneity were developed in Stata Software (Stata/MP 14), but as the gamma frailty models fitted well when compared to clus-
tered heterogeneity (based on the AIC criterion: Akaike, 1973), the results of Weibull AFT models with gamma frailty are
presented and discussed further. The models were developed by considering all the explanatory variables mentioned in
Table 2, except for the variables: Traffic offence due to phone use and Accidents. As none of the drivers responded any crash
history or traffic offences in past three years due to phone use, so the self-reported crash history was not taken into account
while developing the models. Tables 4 and 5 show the summary (including the parameter estimates, standard error, etc.) of
Weibull AFT (with gamma frailty) models for driver’s responses to the pedestrian crossing event and the road crossing event
by parked vehicles respectively. Parameters showing the goodness of fit of the model, are mentioned at the bottom of both
the Tables. For the developed models, p (scale parameter) estimates were 2.97 and 2.63 for the pedestrian crossing event and
the road crossing event by parked vehicles respectively. The scale parameters for both the models are greater than 1 and this
shows that probability of detecting the event got increased with elapsed time for both the events. Intercepts of the models
possess all the attributes of reference categories of categorical variables and mean value of continuous variables. Thus in a
broader sense, intercepts are showing the reaction time behaviour of young drivers in non-distracted driving conditions. Col-
umn exp (b) shows the percentage change in reaction time in both the tables.

5.3.1. Results for pedestrian crossing event

In case of pedestrian crossing event, all the four phone distractions: simple conversation, complex conversation, simple
texting and complex texting were found to be significant factors for increasing the reaction times (Table 4). In case of cog-
nitive distraction, both simple and complex difficulty levels caused an increment of 40% and 95% respectively in reaction
times of the drivers. This increment may be attributed to the fact that the drivers, who are involved in talking during driving,
reduce the visual scanning of the peripheral environment; but, focus more at the roadway for continuous longer durations in
order to compensate for the increased workload. This situation might have delayed the driver’s response to detect the event
which started from sides of the road but remained undetected by the drivers until the event proceeded and reached on the
roadway. In case of visual distraction, simple texting caused 137% increment in reaction times while complex texting caused
204% increment. For performing visual distraction tasks concurrently with the driving, the drivers reduced the frequency of
visual scanning of roadway as well as peripheral environment, while increased the frequency of observing the phone screen,
which might have led to slow response to the events. Though there is small decrement in reaction time against the pedes-
trian crossing event (at 10% significance level) for the drivers who converse on phone frequently compared to the drivers
who rarely use phone during driving, but, the factor talking during driving did not come out to be significant at 5% significant
level.

Table 4
Summary statistics of Weibull AFT model developed for reaction times for pedestrian crossing event.

Variables Coefficient Exp(b) Standard error z-value p-value [95% Confidence

Interval]
Simple conversation 0.339 1.40 0.073 4.64 <0.001 0.196 0.482
Complex conversation 0.669 1.95 0.089 7.52 <0.001 0.495 0.844
Simple texting 0.863 2.37 0.076 11.41 <0.001 0.715 1.012
Complex texting 1.115 3.04 0.092 12.09 <0.001 0.934 1.296
Frequently conversing 0.135 0.87 0.077 1.76 0.079 0.286 0.016
Less frequently conversing 0.083 0.92 0.084 0.99 0.322 0.248 0.081
Intercept 0.067 0.071 0.94 0.345 0.072 0.206
Variance of gamma frailty 0.369 0.157 0.16 0.847

Chi-square = 155.62 on 6 degrees of freedom, p-value < 0.001.

Likelihood ratio test of h: Chi-square value = 18.34, p- value < 0 0.001.
Number of observations = 250.
Number of groups = 100.
p (scale parameter) = 2.97.
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 361

Table 5
Summary statistics of Weibull AFT model developed for reaction times for road crossing event by parked vehicles.

Variables Coefficient Exp (b) Standard error z-Value p-Value [95% Confidence
Interval]
Simple Conversation 0.395 1.48 0.083 4.760 <0.001 0.232 0.557
Complex Conversation 0.504 1.65 0.092 5.450 <0.001 0.323 0.685
Simple Texting 0.796 2.21 0.093 8.600 <0.001 0.615 0.977
Complex Texting 0.997 2.71 0.084 11.860 <0.001 0.832 1.162
Frequently Conversing 0.209 0.81 0.078 2.700 0.007 0.361 0.057
Less Frequently Conversing 0.177 0.83 0.086 2.060 0.039 0.345 0.009
Intercept 0.214 0.086 2.490 0.013 0.046 0.382
Variance of gamma frailty 0.252 0.132 0.090 0.704

Chi-square = 128.01 on 6 degrees of freedom, p-value < 0.001.

Likelihood ratio test of h: Chi-square value = 11.24, p-value < 0.001.
Number of observations = 250.
Number of groups = 100.
p (scale parameter) = 2.63.

5.3.2. Results for road crossing event by parked vehicles

Similar to pedestrian crossing event, all the phone use conditions were significant in case of the road crossing event by
parked vehicles (Table 5). The observed increment in reaction time was 48%, 65%, 121% and 171% due the presence of simple
conversation, complex conversation, simple texting and complex texting respectively. The reason for increment in reaction
time during distracted drivers could be similar to the pedestrian crossing event; however, the increments observed in the
vehicle crossing event are little lesser than the increments observed in case of pedestrian crossing event. The reason could
that the driver found pedestrian crossing event as more sudden compared to the road crossing event by parked vehicles, and
thus took more time to perceive the event. The variable ‘talking during driving’ also showed significant effect on reaction
times of the drivers in case of the road crossing event by parked vehicles. Participants who had a habit of receiving all
the calls during driving (frequently conversing) or receiving one or two calls per day during driving (less frequently conversing)
had 19% and 17% lesser reaction times respectively, than the participants who rarely use phone while driving. More expe-
rience of mobile phone use during driving, might have enabled them to perform slightly better than the others.
The increments due to simple conversation in both the event cases are similar to the previous studies (Patten et al., 2004;
Haque and Washington, 2014); but, in complex conditions the results are much more than previously stated results; here
one of the reasons could be that increased share of attention was allocated to the secondary task rather than driving in order
to handle the increased level of complexity in conversation. The other possible reasons could be the type of event used for
testing and different driving behaviour of Indian drivers. In previous studies also, the difference in reaction times due to dif-
ferent event and driver behaviour can be seen. For example, Beede and Kass (2006) measured reaction time against a signal
detection task and the mean reaction time was 1.24 s in no phone condition and 1.33 s in with phone condition. The mean
reaction time observed against the Peripheral Detection Task by Patten et al. (2004) during no phone use, simple conversa-
tion and complex conversation were 0.584 s, 0.656 s and 0.845 s respectively. Thus from the earlier studies, it can be seen
that reaction time varies for different events, and for present study also some variation from previous studies were expected.
The present study results showed that complex conversation, simple and complex texting caused much more increment in
Indian drivers’ reaction time when compared to previous studies. The driver behaviour is also reflected through their phone
use habits during driving which can be accounted by including these habits as explanatory variables. The present study
results showed the phone use habits significantly altered driver’s response behaviour in case of the road crossing event
by parked vehicles. However, in case of the pedestrian crossing event the phone use habits did not have any significant effect
on reaction times.
In both the models, driver demographic (age and gender) factors did not exhibit any significant effect on reaction times. In
some of the previous studies, it was identified that reaction time behaviour deteriorates more for older drivers compared to
younger drivers (Hancock et al., 2003; Caird et al., 2008). However, Oviedo-Trespalacios et al. (2016) conducted a critical and
systematic review of existing research on the impacts of mobile phone distraction and observed that many previous studies
have reported a non-significant effect of age and gender differences on various driving performance measures. Other factors
such as: phone type, trip types, driving experience and texting during driving also did not prove to be significantly effective in
changing the reaction times for both the events.

5.3.3. Survival rates

Graphical representation of the increment/decrement in the probabilities of not detecting the events (survival rates) due
to the significant factors could be calculated with the help of parameter estimates of the developed models by using the sur-
vival function described in Eq. (5). For example, to find out the probability of not detecting the pedestrian crossing event at
1 s in simple conversation task, value of X is kept equal to 1 for simple conversation category in the equation and is kept 0 for
all the other categories of phone use conditions, and value of t is kept as 1. Following is the calculation for simple conver-
sation task for two different time steps, t = 1 and t = 2.
362 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

Sðt ¼ 1Þ ¼ expððexpð0:067  :339Þ  1Þ2 Þ

Sðt ¼ 1Þ ¼ 0:641

Sðt ¼ 2Þ ¼ expððexpð0:067  :339Þ  2Þ2 Þ

Sðt ¼ 2Þ ¼ 0:169
Likewise the predicted probabilities of not detecting the events for all the phone use conditions, for young drivers includ-
ing all attributes of reference categories are shown in Fig. 6 (for both the events).
As anticipated the probabilities of not detecting the event got decreased as the time elapsed i.e. the chances of detecting
the event increased with the increase in elapsed time of the event. The survival rates were higher for complex conversation
tasks at the same time step (e.g., at 2 s) than for the simple conversation task and no-phone use conditions (for both the
events). And minimum survival rates were observed for no phone use condition in both the event cases.
Fig. 7 shows difference between survival rates in case of road crossing event by parked vehicles, for two different groups:
drivers who use phone frequently while driving and drivers who rarely use phone (or do not use it at all) during driving. The
survival rates for frequent mobile user during driving are lesser compared to infrequent users in all phone use conditions, i.e.

Fig. 6. Probabilities of not detecting (A) Pedestrian crossing event (B) Road crossing event by parked vehicles in all phone use conditions.
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 363

Fig. 7. Comparisons of probabilities of not detecting the road crossing event by parked vehicles for (A) frequent and (B) infrequent users of mobile phone
during driving in cognitively distracted driving.

the effect of phone use is reduced for the frequent phone users as their driving behaviour have been altered due to the habit
of using phone during driving.

6. Conclusions

Present study analysed the distraction effects of texting and talking on mobile phone during driving by taking reaction
time as the performance measure. Reaction times were measured against the simulated events (pedestrian crossing and
parked vehicles entering the road) which were similar to some of the real world events in Indian context. In preliminary
analysis of reaction time data the 90th percentile value of reaction times found to be more than 2.5 s (standard value set
by AASHTO and adopted by IRC) in all distracted driving conditions (except for simple conversation task). Statistical mod-
elling of the adverse effects of mobile phone distraction on reaction times was also done by developing Weibull AFT models
for both the events. Factors, other than distraction (such as: driver demographics, driving characteristics, mobile phone use
habits in everyday life and during driving) were also taken into consideration while developing the models. The results of
the models for both the events showed that presence of any kind of mobile phone distraction was the main reason for
impairment in driver’s response behaviour. The presence of simple conversation, complex conversation, simple texting
and complex texting caused 40%, 95%, 137% and 204% increment of reaction times of drivers in case of pedestrian crossing
event; and similarly, caused 48%, 65%, 121% and 171% increment of reaction times in case of road crossing event by parked
vehicles respectively. The model results indicated that in both the event cases difficulty level of the task also affected the
364 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365

responding behaviour of the drivers. Here, it can be speculated that the reduced responsiveness to road side events is an out-
come of increased competition between attention resources while performing secondary tasks during driving. There was a
significant difference in the survival rates (or the predicted probabilities that the events remained undetected by the drivers)
of no phone use driving conditions and distracted driving conditions due to phone usages during driving. Age and gender
factors did not prove to be significant in this study.
Driver’s reaction time is one of the main factors to decide whether the crash can be avoided or not. Therefore, the algo-
rithm used by many ITS applications (e.g., collision warning systems) is mainly dependent on reaction times. Some threshold
value of reaction time ranging from 0.7 to 1.5 s is mostly used in these algorithms (Green, 2000; Sultan and McDonald, 2003;
Doecke et al., 2015). The results from this study show that even if the drivers are warned 1.5 s ahead of the hazard, then also
36%, 59%, 70%, 80% of the drivers (based on the survival rates for pedestrian crossing event) will have insufficient time to
react to the hazard if they are involved in simple conversation, complex conversation, simple texting or complex texting task
during driving respectively. According to the present study, in India, where more than half of the drivers use phone during
driving (for conversation purposes), the adoption of such threshold reaction time value for warning systems may even result
in more collisions.
As per the rules and regulations by the Motor Vehicle Act (MVA) (1988) and Central Motor Vehicle Rules (1989) use of
mobile phone while driving is considered as an offence in India. But the lack of strict enforcement of these rules, the drivers
are continuing the mobile phone use during driving without knowing its severe impact on driving (Urie et al., 2016). There-
fore, the statistical results/proofs of the present study can help the regulatory bodies of developing countries, to spread the
public awareness about the serious negative impacts associated with phone use during driving.

6.1. Future work

In this study only reaction time was taken as a performance measure, therefore future scope could be to investigate the
distraction effects on other measures such as lane deviation and speed change. In the present study, a driving simulator was
used which might produce different results compared to real-world driving so the validation of the study in field is required.
Moreover, the sample sizes for older drivers (compared to young and mid-age drivers) and female drivers (compared to male
drivers) were small; therefore, there is a scope of improving the analysis by increasing the sample size of these categories.
Further, in the present study only simple rural road geometry is incorporated in the scenario, which might not truly describe
overall reaction time behaviour of the drivers in Indian conditions; therefore, future research work could be to include com-
plex urban road geometries in the simulated scenarios and compare the findings for urban and rural environments. The dis-
traction studies often indicate the link between accidents and phone use during driving but true scale of the link between the
two is still not available; so further attempts should be made to get the more precise relationship between phone use and
corresponding accidents.

References

AASHTO (American Association of State Highway and Transportation Officials), 2011. Policy on Geometric Design of Highways and Streets, 2011.
Akaike, H., 1973. Information theory and an extension of the maximum likelihood principle. In: Petrov, B.N., Csaki, F. (Eds.), Proceedings of the Second
International Symposium on Information Theory. Akademiai Kiado, Budapest, pp. 267–281.
Al-Darrab, I.A., Khan, Z.A., Ishrat, S.I., 2009. An experimental study on the effect of mobile phone conversation on drivers’ reaction time in braking response.
J. Safety Res. 40 (3), 185–189.
Alm, H., Nilsson, L., 1994. Changes in driver behaviour as a function of hands-free mobile phones—a simulator study. Accid. Anal. Prev. 26 (4), 441–451.
Alm, H., Nilsson, L., 1995. The effects of a mobile telephone task on driver behaviour in a car following situation. Accid. Anal. Prev. 27 (5), 707–715.
Ascone, D., Lindsey, T., Varghese, C., 2009. Traffic Safety Factor: An Examination of Driver Distraction as Recorded in NHTSA Databases. National Highway
Traffic Safety Administration’s National Center for Statistical and Analysis.
Beede, K.E., Kass, S.J., 2006. Engrossed in conversation: the impact of cell phones on simulated driving performance. Accid. Anal. Prev. 38 (2), 415–421.
Bellinger, D.B., Budde, B.M., Machida, M., Richardson, G.B., Berg, W.P., 2009. The effect of cellular telephone conversation and music listening on response
time in braking. Transp. Res. Part F: Traffic Psychol. Behav. 12 (6), 441–451.
Benedetto, A., Calvi, A., D’Amico, F., 2012. Effects of mobile telephone tasks on driving performance: a driving simulator study. Adv. Transp. Stud. 26, 29–44.
Brookhuis, K.A., de Vries, G., de Waard, D., 1991. The effects of mobile telephoning on driving performance. Accid. Anal. Prev. 23 (4), 309–316.
Burns, P.C., Parkes, A., Burton, S., Smith, R.K., Burch, D., 2002. How Dangerous Is Driving With a Mobile Phone?: Benchmarking the Impairment to Alcohol.
Transport Research Laboratory, Berkshire, United Kingdom.
Burns, P., Lécuyer, J.F., Chouinard, A., 2008, June. Observed driver phone use rates in Canada. In: Proceedings of the 18th Canadian Multidisciplinary Road
Safety Conference, Whistler, British Columbia.
Caird, J.K., Johnston, K.A., Willness, C.R., Asbridge, M., Steel, P., 2014. A meta-analysis of the effects of texting on driving. Accident Anal. Prev. 71, 311–318.
Caird, J.K., Willness, C.R., Steel, P., Scialfa, C., 2008. A meta-analysis of the effects of cell phones on driver performance. Accid. Anal. Prev. 40 (4), 1282–1293.
Central Motor Vehicle Rules, 1989 Available at <http://www.tn.gov.in/sta/Cmvr1989.pdf>. accessed on 1-07-2016.
Choudhary, P., Velaga, N.R., 2017. Analysis of vehicle-based lateral performance measures during distracted driving due to phone use. Transp. Res. Part F:
Traffic Psychol. Behav. 44, 120–133.
Cleves, M., 2008. An Introduction to Survival Analysis Using Stata. Stata Press, College Station, Texas.
Consiglio, W., Driscoll, P., Witte, M., Berg, W.P., 2003. Effect of cellular telephone conversations and other potential interference on reaction time in a
braking response. Accid. Anal. Prev. 35 (4), 495–500.
Cooper, J., Yager, C., Chrysler, S.T., 2011. An Investigation of the Effects of Reading and Writing Text-Based Messages While Driving (Report No. SWUTC/11/
476660-00024-1).
DGT, 2008. Anuario Estadístico de Accidentes de la Dirección General de Tráfico, 2008 (Statistical Yearbook of Accidents of the Directorate-General for
Traffic). Spain.
Doecke, S., Grant, A., Anderson, R.W., 2015. The real-world safety potential of connected vehicle technology. Traffic Injury Prevent. 16 (sup1), S31–S35.
Eby, D.W., Vivoda, J.M., Louis, R.M.S., 2006. Driver hand-held cellular phone use: a four-year analysis. J. Safety Res. 37 (3), 261–265.
 P. Choudhary, N.R. Velaga / Transportation Research Part C 77 (2017) 351–365 365

Fambro, D.B., Fitzpatrick, K., Koppa, R.J., 1997. National Cooperative Highway Research Program Report 400: Driver Perception-Brake Response in Stopping
Sight Distance Situations. NCHRP, Transportation Research Record: Journal of the Transportation Research Board, Washington, DC.
Green, M., 2000. ‘‘How Long Does It Take to Stop?” Methodological analysis of driver perception-brake times. Transp. Human Factors 2 (3), 195–216.
Gutierrez, R.G., 2002. Parametric frailty and shared frailty survival models. Stata J. 2 (1), 22–44.
Hancock, P.A., Lesch, M., Simmons, L., 2003. The distraction effects of phone use during a crucial driving maneuver. Accid. Anal. Prev. 35 (4), 501–514.
Haque, M.M., Washington, S., 2014. A parametric duration model of the reaction times of drivers distracted by mobile phone conversations. Accid. Anal.
Prev. 62, 42–53.
Haque, M.M., Washington, S., 2015. The impact of mobile phone distraction on the braking behaviour of young drivers: a hazard-based duration model.
Transp. Res. Part C: Emerg. Technol. 50, 13–27.
Harms, L., Patten, C., 2003. Peripheral detection as a measure of driver distraction. A study of memory-based versus system-based navigation in a built-up
area. Transp. Res. Part F: Traffic Psychol. Behav. 6 (1), 23–36.
Horberry, T., Anderson, J., Regan, M.A., Triggs, T.J., Brown, J., 2006. Driver distraction: the effects of concurrent in-vehicle tasks, road environment complexity
and age on driving performance. Accid. Anal. Prev. 38 (1), 185–191.
Horrey, W.J., Wickens, C.D., 2006. Examining the impact of cell phone conversations on driving using meta-analytic techniques. Human Fact.: J. Human Fact.
Ergon. Soc. 48 (1), 196–205.
Huth, V., Sanchez, Y., Brusque, C., 2015. Drivers’ phone use at red traffic lights: A roadside observation study comparing calls and visual–manual
interactions. Accident Anal. Prev. 74, 42–48.
IRC 66: 1976 – Recommended Practices for Sight Distance on Rural Highways.
Jamson, A.H., Merat, N., 2005. Surrogate in-vehicle information systems and driver behaviour: effects of visual and cognitive load in simulated rural driving.
Transp. Res. Part F: Traffic Psychol. Behav. 8 (2), 79–96.
Johansson, G., Rumar, K., 1971. Drivers’ brake reaction times. Human Fact.: J. Human Fact. Ergon. Soc. 13 (1), 23–27.
Klauer, S.G., Dingus, T.A., Neale, V.L., Sudweeks, J.D., Ramsey, D.J., 2006. The Impact of Driver Inattention on Near-Crash/Crash Risk: An Analysis Using the
100-Car Naturalistic Driving Study Data (No. HS-810 594).
Lamble, D., Kauranen, T., Laakso, M., Summala, H., 1999. Cognitive load and detection thresholds in car following situations: safety implications for using
mobile (cellular) telephones while driving. Accid. Anal. Prev. 31 (6), 617–623.
Lee, J.D., Caven, B., Haake, S., Brown, T.L., 2001. Speech-based interaction with in-vehicle computers: The effect of speech-based e-mail on drivers’ attention
to the roadway. Human Fact.: J. Human Fact. Ergon. Soc. 43 (4), 631–640.
Lee, J.D., Young, K.L., Regan, M.A., 2008. Defining driver distraction. Driver Distraction: Theory Effects Mitigation 13 (4), 31–40.
Leung, S., Croft, R.J., Jackson, M.L., Howard, M.E., Mckenzie, R.J., 2012. A comparison of the effect of mobile phone use and alcohol consumption on driving
simulation performance. Traffic Injury Prevent. 13 (6), 566–574.
Massachusetts Institute of Technology, 1935. Report of the Massachusetts Highway Accident Survey CWA and ERA Project. Massachusetts Institute of
Technology, MA.
Matthews, R., Legg, S., Charlton, S., 2003. The effect of cell phone type on drivers subjective workload during concurrent driving and conversing. Accid. Anal.
Prev. 35 (4), 451–457.
McEvoy, S., Stevenson, M., 2007. An exploration of the role of driver distraction in serious road crashes Distracted Driving. Australasian College of Road
Safety, Sydney, pp. 189–211.
McKnight, A.J., McKnight, A.S., 1993. The effect of cellular phone use upon driver attention. Accid. Anal. Prev. 25 (3), 259–265.
MRTH, 2013. Road Accidents in India, 2013. Transport Research Wing (TRW) of Ministry of Road Transport & Highways, Government of India.NCDB, 2007.
National Collision Database. Transport Canada.
Motor Vehicle Act (MVA), 1988. Available at <http://www.tn.gov.in/sta/Mvact1988.pdf> (accessed on 1-07-2016).
Normann, O.K., 1953. Braking Distances of Vehicles from High Speeds, Proceedings of the Highway Research Board, 22, Highway Research Board,
Washington, DC, pp. 421–436.
NRSC, 2010. Safer Journeys. New Zealand’s road safety strategy 2010. National Road Safety Committee, Ministry of Transport.
Oviedo-Trespalacios, O., Haque, M.M., King, M., Washington, S., 2016. Understanding the impacts of mobile phone distraction on driving performance: a
systematic review. Transp. Res. Part C: Emerg. Technol. 72, 360–380.
Patten, C.J., Kircher, A., Östlund, J., Nilsson, L., 2004. Using mobile telephones: cognitive workload and attention resource allocation. Accid. Anal. Prev. 36 (3),
341–350.
Peng, Y., Boyle, L.N., Lee, J.D., 2014. Reading, typing, and driving: how interactions with in-vehicle systems degrade driving performance. Transp. Res. Part F:
Traffic Psychol. Behav. 27, 182–191.
Prat, F., Planes, M., Gras, M.E., Sullman, M.J.M., 2015. An observational study of driving distractions on urban roads in Spain. Accid. Anal. Prev. 74, 8–16.
Recarte, M.A., Nunes, L.M., 2003. Mental workload while driving: effects on visual search, discrimination, and decision making. J. Exp. Psychol.: Appl. 9 (2),
119–137.
Redelmeier, D.A., Tibshirani, R.J., 1997. Association between cellular-telephone calls and motor vehicle collisions. N. Engl. J. Med. 336 (7), 453–458.
Santos, J., Merat, N., Mouta, S., Brookhuis, K., De Waard, D., 2005. The interaction between driving and in-vehicle information systems: comparison of results
from laboratory, simulator and real-world studies. Transp. Res. Part F: Traffic Psychol. Behav. 8 (2), 135–146.
Shabeer, H., Banu, W., 2012. Mobile phone accidents-experience of India. Transp. Telecommun. 13 (3), 193–208.
Strayer, D.L., Drew, F.A., 2004. Profiles in driver distraction: effects of cell phone conversations on younger and older drivers. Human Fact.: J. Human Fact.
Ergon. Soc. 46 (4), 640–649.
Sullman, M.J., Prat, F., Tasci, D.K., 2015. A roadside study of observable driver distractions. Traffic Injury Prevent. 16 (6), 552–557.
Sultan, B., McDonald, M., 2003. Assessing the safety benefit of automatic collision avoidance systems (during emergency braking situations). In: Proceedings
of the 18th International Technical Conference on the Enhanced Safety of Vehicle. (DOT HS 809 543).
Thulin, H., Gustafsson, S., 2004. Mobile Phone Use While Driving. Swedish National Road and Transport Research Institute, Linköping.
Törnros, J., Bolling, A., 2006. Mobile phone use–effects of conversation on mental workload and driving speed in rural and urban environments. Transp. Res.
Part F: Traffic Psychol. Behav. 9 (4), 298–306.
Urie, Y., Velaga, N.R., Maji, A., 2016. Cross-Sectional Study of Road Accidents and Related Law Enforcement Efficiency for Ten Countries: A Gap Coherence
Analysis. Traffic Inj. Prev. 17 (7), 686–691. http://dx.doi.org/10.1080/15389588.2016.1146823.
Washington, S. P., Karlaftis, M. G., Mannering, F. L., 2011. Statistical and econometric methods for transportation data analysis.
WHO fact sheets, May 2014: The Top 10 Causes of Death, available at <http://www.who.int/mediacentre/factsheets/fs310/en/> (accessed on 31-01-2016).
World Health Organization, 2011. Mobile Phone Use: A Growing Problem of Driver Distraction.
Yannis, G., Laiou, A., Papantoniou, P., Christoforou, C., 2014. Impact of texting on young drivers’ behavior and safety on urban and rural roads through a
simulation experiment. J. Safety Res. 49 (2014), 25–31.
Young, K.L., Rudin-Brown, C.M., Lenné, M.G., 2010. Look who’s talking! a roadside survey of drivers’ cell phone use. Traffic Injury Prevent. 11 (6), 555–560.