Content uploaded by Nick Reed
Author content
All content in this area was uploaded by Nick Reed on Sep 26, 2014
Content may be subject to copyright.
Transport Research Laboratory
The Effect Of Text Messaging On Driver
Behaviour
A Simulator Study
by N. Reed & R. Robbins
PPR 367
PUBLISHED PROJECT REPORT
Transport Research Laboratory
PUBLISHED PROJECT REPORT PPR 367
The Effect Of Text Messaging On Driver Behaviour
A Simulator Study
by N. Reed & R. Robbins (TRL)
Prepared for: Project Record: Texting Whilst Driving
Client: RAC Foundation
(Elizabeth Dainton)
Copyright Transport Research Laboratory September 2008
This Published Report has been prepared for RAC Foundation. Published Project Reports
are written primarily for the Client rather than for a general audience and are published
with the Client’s approval.
The information contained herein is the property of TRL Limited and the views expressed
are those of the author(s) and not necessarily those of RAC Foundation. Whilst every
effort has been made to ensure that the matter presented in this report is relevant,
accurate and up-to-date at the time of publication, TRL Limited cannot accept any
liability for any error or omission.
Name Date
Approved
Project
Manager Nick Reed 09/09/2008
Technical
Referee Andrew Parkes 10/09/2008
Published Project Report
TRL PPR 367
If this report has been received in hard copy from TRL, then in support of the company’s
environmental goals, it will have been printed on recycled paper, comprising 100% post-
consumer waste, manufactured using a TCF (totally chlorine free) process.
Published Project Report
TRL
i
PPR 367
Contents
List of Figures iv
List of Tables vi
Executive summary viii
Abstract 1
1
Introduction 3
2
Method 6
2.1
Participants 6
2.2
Study design 6
2.3
Equipment 6
2.4
Familiarisation 7
2.5
Participant instructions 7
2.6
Questionnaires 8
2.7
Route design 9
2.7.1
Reaction time events 10
2.7.2
Texting tasks 10
2.7.3
Overall design 13
2.8
Trial procedure 14
2.9
Recorded simulator data 15
2.10
Calculation 15
3
Results 16
3.1
Participants 16
3.2
Reaction time (RT) tasks 16
3.2.1
Response rate 16
3.2.2
Reaction times 16
3.3
Analyses by section 17
3.3.1
Speed 17
3.3.2
Variation in lane position 19
3.4
Analyses by texting episode 20
3.4.1
Section 1 (Motorway 1) 21
3.4.2
Section 2 (Loops) 21
3.4.3
Section 3 (Car following) 23
3.4.4
Section 4 (Motorway 2) 27
3.4.5
Texting completion times 28
3.5
Gender differences 28
3.6
Error rate 29
3.7
Patterns of mobile phone use 30
3.7.1
Familiarity with mobile phones and text messaging 30
Published Project Report
TRL
ii
PPR 367
3.7.2
Storage and method of use of mobile phones whilst driving 31
3.7.3
Baseline texting completion time 32
3.7.4
Comparison of text completion time when driving vs. baseline 33
3.8
Perceptions of the legality of mobile phone use while driving 34
3.9
Perceptions of the relative risks of driving behaviours 35
3.10
Subjective effects of Texting on performance 36
3.10.1
Recall of text messages received when driving 36
3.10.2
Driving performance in the Texting and Control drives 37
3.10.3
Differences in performance when sending or receiving a text
message 39
3.11
Personality tests 43
4
Discussion 44
4.1
Comparison with previous distraction studies 45
4.1.1
Reaction times 46
4.1.2
Speed 46
4.1.3
SDLP 46
Acknowledgements 48
References 48
Appendix A
The TRL Driving Simulator Centre 50
Appendix B
Phone makes and network providers 53
Appendix C
Messages sent in Texting drive 54
Published Project Report
TRL
iii
PPR 367
Published Project Report
TRL
iv
PPR 367
List of Figures
Figure 2.1 Car-following route with motorway chevrons ............................................. 7
Figure 2.2 The trigger stimulus for the visual reaction time event.............................. 10
Figure 2.3 Overall design of Texting drive .............................................................. 13
Figure 3.1 Mean reaction times to each of the RT tasks............................................ 17
Figure 3.2 The mean of Maximum speed values observed across participants in each
drive ............................................................................................................ 18
Figure 3.3 The mean speed values observed across participants in Section 3 of each
drive ............................................................................................................ 19
Figure 3.4 Mean SDLP across participants observed in section 3................................ 20
Figure 3.5 Mean SDLP in the Write 2 task relative to that observed in the Control drive 22
Figure 3.6 Total number of lane departures observed in the Write 2 task compared to
those in the Control drive ............................................................................... 22
Figure 3.7 Total number of lane departures observed in the Read 2 task compared to
those in the Control drive ............................................................................... 23
Figure 3.8 Mean speed and Maximum speed observed in the Write 3 task relative to that
observed in the Control drive .......................................................................... 24
Figure 3.9 Mean, Standard Deviation of, and Minimum time headway relative to the lead
vehicle for the Write 3 task during the car following section relative to that observed
in the Control drive ........................................................................................ 25
Figure 3.10 Mean SDLP in the Write 3 task relative to that observed in the Control drive
................................................................................................................... 26
Figure 3.11 Total number of lane departures observed in the Write 3 task compared to
those in the Control drive ............................................................................... 27
Figure 3.12 Text completion times when driving...................................................... 28
Figure 3.13 Plots of the interaction between gender and (a) SDLP in Write 2 and (b)
Mean speed in Write 3.................................................................................... 29
Figure 3.14 Weekly use of mobile phone for spoken conversations and text messaging.30
Figure 3.15 Participants’ ratings of ease of use of their mobile phones ....................... 31
Figure 3.16 Frequency with which participants leave their mobile phones on silent ...... 32
Figure 3.17 Frequency with which participants use their mobile phones 'hands-free' .... 32
Figure 3.18 Distribution of participants’ mean baseline texting completion times ......... 33
Figure 3.19 Text completion times when driving compared to the baseline condition ... 33
Figure 3.20 Participants’ understanding of the current legality of mobile phone use ..... 34
Figure 3.21 Participants’ beliefs about whether various behaviours should be legal ...... 35
Figure 3.22 Participants’ mean ratings of performance effects of sending or receiving a
text message on Concentration and Keeping in lane........................................... 41
Figure 3.23 Participants’ mean ratings of performance effects of sending or receiving a
text message on Speed, Distance, Hazard awareness and General performance .... 42
Figure B.4.1 Count of (a) mobile phone makes and (b) network providers used in the
study ........................................................................................................... 53
Published Project Report
TRL
v
PPR 367
Published Project Report
TRL
vi
PPR 367
List of Tables
Table 2.1 IPIP facets used and their hypothesised relation to texting whilst driving ....... 9
Table 2.2 The road sections used for the simulator trial .............................................9
Table 2.3 Text messages composed by participants whilst driving ............................. 11
Table 2.4 Text messages received by participants whilst driving................................ 11
Table 2.5 Text message to be ignored by participants whilst driving .......................... 11
Table 2.6 Timed text messages composed by participants without distraction ............. 12
Table 2.7 Trial procedure ..................................................................................... 14
Table 2.8 Data recorded by the simulator at 20Hz ................................................... 15
Table 3.1 A ranked list of the mean risk ratings assigned to driving behaviours........... 36
Table 3.2 Post-trial message recall performance ..................................................... 36
Table 3.3 Descriptive statistics for participants’ perceptions of their performance in the
Texting and Control drives (* Ratings represent percentages where high values
correspond to superior driving performance). .................................................... 38
Table 3.4 Differences in participants’ perceptions of their performance in the Texting and
Control drives (paired samples t-tests)............................................................. 39
Table 3.5 Differences in participants’ perceptions of performance impairment when
sending or receiving a text message. ............................................................... 40
Table C.1 Text messages composed by each participant in their Texting drive ............ 54
Published Project Report
TRL
vii
PPR 367
Published Project Report
TRL
viii
PPR 367
Executive summary
The detrimental effects of mobile phone use on driver performance have been widely
studied (e.g. Brown & Poulton, 1961; Burns, Parkes, Burton, Smith & Burch, 2002;
Parkes, Luke, Burns & Lansdown, 2007). However, there is a significant gap in the
research literature, namely the effects of text messaging on driving. In 2008, The RAC
Foundation asked 2002 members of the social networking website Facebook
(www.facebook.com), to self report on whether they text whilst driving and 45%
admitted doing so. In response to this, the RAC Foundation commissioned TRL to
investigate the relative driver impairment caused by texting whilst driving.
The aim of the study was to assess the impact of text messaging on driver performance,
and the attitudes and beliefs that surrounded the activity in the 17-25 age category.
Reaction times, car following ability, lane control, and driver speed were used as
measures of driver performance and driver attitudes were assessed using personality
questionnaires. The TRL driving simulator was used to conduct the research. Seventeen
participants between the ages of 17-24 were recruited for the study (8 male; 9 female).
All participants described themselves as regular users of text messaging and used
phones with standard key pads (i.e. alphanumeric key pads. Other phone types were
excluded). It was hypothesised that when writing/reading text messages, drivers would
display increased reaction times, poorer car following ability, poorer lateral lane control,
and reduced speed. It was also hypothesised that reductions in drivers’ performance will
be greater when writing a text message than when reading a text message.
Participants first completed a familiarisation drive, which consisted of a ten minute
motorway drive in which the participant was required to follow a lead vehicle at a safe
distance designated with chevrons. This was followed by two identical test drives. In one
of the drives, participants were required to complete text messaging tasks following
verbal instructions (read a received message; compose and send a message to a
contact; ignore an incoming message). In the other drive, participants completed the
same route without any distractions. Questionnaires were completed before and after
the simulator drives. Participants were also timed completing a comparable set of text
messages to those used in the simulator drive to investigate how much longer it takes to
text when driving.
Results demonstrated that participants’ driving behaviour was impaired by concurrent
text message tasks. Writing text messages created a significantly greater impairment
than reading text messages. Behaviour in response to the arrival of an ignored text
message was unaffected. Reaction times to (task-unrelated) trigger stimuli tended to be
higher when reading or writing a message. The slowest average reaction time was
observed for drivers responding to the visual reaction time task whilst trying to compose
a text message where reaction times increased from 1.2 to 1.6 seconds. Furthermore,
participants were significantly more likely to fail to respond to the reaction time stimuli if
engaged in concurrent text messaging whilst text completion times when driving were
nearly three times longer than when composing similar messages undistracted. The
failure to detect hazards, increased response times to hazards, and exposure time to
that risk have clear implications for safety. At motorway speeds (as were present in the
visual RT task), a driver would travel more than one mile whilst completing the text
message and the increase in mean reaction time would result in an increased stopping
distance of 12.5m (approximately three car lengths). This could easily make the
difference between causing and avoiding an accident or between a fatal and non-fatal
collision.
It was observed that drivers tended to reduce their speed in the texting conditions. It is
suggested that drivers were aware that their driving was impaired to some degree whilst
engaged in text messaging tasks and chose to reduce their speed in order to mitigate
accident risk. The most conspicuous change in performance was the large increases in
variability of lane position resulting in many more lane departures when texting. It was
further identified that the impairment caused by texting was far more significant for
Published Project Report
TRL
ix
PPR 367
female rather than male drivers. The survey by the RAC Foundation (2008) found that
male drivers were more likely to text and drive. It is concluded that, although male
drivers may show a reduced impairment when texting and driving, the increased
probability that male drivers will engage in this behaviour suggests that the overall
impairment across the sexes may be more equal.
It was observed that participants, when texting, were less able to maintain a constant
distance behind a lead vehicle and showed increased variability in lateral lane position
when following that vehicle. In real world traffic situations, it is suggested that poorer
control of vehicle speed, lateral position, and increased reaction times in this situation
would increase the likelihood of collision dramatically.
As hypothesised, reading text messages had a less detrimental effect on performance
than writing messages but a detrimental effect nevertheless. Ignored text messages
appeared to have a negligible effect on performance. This pattern of results is consistent
with a lower relative task demand of reading a text message compared to writing a
message where, in addition to viewing the phone display screen, the driver must
consider the text to be written and interact with the phone to compose the message.
The questionnaire results indicated that participants were confused about the legality of
texting whilst driving. A majority of participants felt that use of a phone for texting whilst
handheld or in a cradle should be illegal. Participants reported feeling impaired in their
driving when texting recognising that they had poorer lane positioning, chose to drive
more slowly, and kept larger safety margins. They also recognised that writing/sending a
message was more of a distraction than reading an incoming message.
Results in the study were compared to three earlier TRL studies that used a similar
methodology (Burns, Parkes, Burton, Smith & Burch, 2002; Sexton, Tunbridge, Brook-
Carter, Jackson, Wright, Stark, & Englehart, 2000; Sexton, Tunbridge, Board, Jackson,
Wright, Stark, & Englehart, 2002. Reaction time impairment caused by texting whilst
driving was apparently greater than that caused by alcohol consumption to the legal limit
for driving, cannabis, and handsfree conversations but less detrimental than using a
mobile phone for handheld conversations. Participants tended to drive more slowly when
texting whilst driving than when conversing on a mobile phone but the speed reduction
was less than observed when drivers were under the influence of cannabis. This suggests
participants feel that they have to compensate for a greater perceived behavioural
impairment caused by texting whilst driving than that caused by talking whilst driving
but less when experiencing the combined physiological and psychological effect of
cannabis. Burns et al. found that there were no significant differences in lateral lane
control in the handheld or handsfree conversation conditions. Driving whilst at the legal
limit of alcohol consumption did result in significantly less steady lane keeping than any
of the other conditions in the study. Sexton et al. (2000) found that drivers displayed an
increase of around 35% in lateral position variability with high does of cannabis whilst
Sexton et al. (2002) found an approximate 14% increase in SDLP for the cannabis and
cannabis + alcohol conditions. This study found that reading messages resulted in a
12.7% increase in lateral position variability whilst that for writing a message increased
by 91.4%. It was further observed that the read and write message tasks were also
accompanied by a significantly greater number of lane departures.
It is concluded that the combination of increased mental workload required to write a
text message, the control impairment caused by the physical act of holding the phone,
and the visual impairment caused by continually shifting visual orientation between the
phone display and the road ahead resulted in significantly impaired ability to maintain
safe road position. Participants’ reduction in speed indicated their awareness of the
impairment caused by texting whilst driving. However, this attempt to mitigate risk
cannot fully compensate for their deterioration in performance when attempting to text
and drive.
Published Project Report
TRL
x
PPR 367
Published Project Report
TRL
xi
PPR 367
Published Project Report
TRL
1
PPR 367
Abstract
RAC Foundation (2008) reported the results of a survey of 2,000+ users of Facebook,
showing that 45% of UK drivers engage in texting whilst driving. The RAC Foundation
commissioned TRL to study the impairment caused by texting whilst driving using TRL’s
driving simulator.
Seventeen drivers (aged 17-24 years) took part in the study. Drivers completed one
drive as normal (undistracted) and one drive in which they completed text messaging
tasks. Participants were impaired in their performance when reading and writing text
messages, particularly reaction time and ability to maintain lateral vehicle control.
Reaction times were around 35% slower when writing a text message. Earlier studies at
TRL showed that alcohol consumption to the legal limit caused a 12% reaction time
increase; cannabis slowed reaction times by 21%. When texting, drivers slowed
significantly, indicating that they recognised the impairment, attempting to mitigate risk
by reducing speed. However, greater lateral variability in lane position and drifting into
adjacent lanes when texting are not mitigated by speed reduction and would lead to
potential conflict with other traffic.
Female drivers showed greater variability in lateral lane position when texting than male
drivers. However, female participants tended to show greater speed reductions indicating
that they may have had greater awareness that their driving was impaired.
This study highlighted that when texting, a driver may present a greater accident risk
than when at the legal limit for alcohol consumption or when under the influence of
cannabis, reinforcing that drivers should refrain from this dangerous activity.
Published Project Report
TRL
2
PPR 367
Published Project Report
TRL
3
PPR 367
1 Introduction
With the emergence of new technologies such as mobile phones, entertainment systems,
in-vehicle information systems (IVIS), etc, the number of distractions to which drivers
are potentially exposed continues to increase. One of the most popular devices used
whilst driving is the mobile phone. Over the last decade mobile phones have transitioned
from a luxury enjoyed by the few, to a must-have item enjoyed by a large proportion of
the world’s population (there are 3.3 billion mobile phone subscriptions worldwide as of
Q4 2007 (The GSM Association, 2007)). There is a clear need to account for the
consequences of mobile phone use on driver performance and behaviour, and a great
deal of research has been conducted into this subject.
Even though the detrimental effects of mobile phone use on driver performance have
been widely studied (Brown & Poulton, 1961; Burns, Parkes, Burton, Smith & Burch,
2002; Strayer, Drews & Crouch, 2006; Horberry, Anderson, Regan, Triggs & Brown,
2006; Parkes, Luke, Burns & Lansdown, 2007; Just, Keller & Cynkar, 2008), there is a
significant gap in the research literature, namely the effects of SMS (Short Message
Service) messaging (texting). To date most research has focused on verbal
communication at best, or at worst conflated text messaging with verbal communication
under vague labels such as ‘mobile phone use’.
Generally, research has failed to recognise that mobile phones are multi-function
devices, and have focused on their primary function, verbal communication. Indeed,
whether verbal communication is still clearly their primary function, especially amongst
young users, is not certain (over 7 billion text messages are sent each day throughout
the world (The GSM Association, 2007)). Despite texting being so popular, its effects on
performance are underrepresented in the research literature. This lack of research is of
concern as a significant number of drivers admit to texting whilst driving (RAC
Foundation, 2008; McEvoy, Stevenson & Woodward, 2006; Gras, Cunill, Sullman, Planes,
Aymerich & Font-Mayolas, 2007; Thulin & Gustafsson, 2004).
In 2008, The RAC Foundation asked 2002 members of the social networking website
Facebook (www.facebook.com), to self report on whether they text whilst driving.
Alarmingly, 45% admitted doing so. They separated those who text into several
categories: 21% read and send message regardless of traffic flows (“multi-tasking
multimedia maestros”), 19% use their phones when stuck in a jam (“opportunistic
optimisers”), and 5% read texts whilst driving but would not respond (“casual
observers”). Research in other countries has also discovered significant amounts of
texting whilst driving (though much less than in the RAC Foundation survey). According
to the McEvoy et al. (2006), 12.4% of Australian drivers admit to having texted whilst
driving. Furthermore, they identify young drivers (18-30) as being significantly more
likely to text whilst driving than older drivers. In similar research, Gras et al. (2006),
found that amongst Spanish drivers, 19.1% admitted texting on highways and 22.5% on
rural roads at least once a month. A survey of Swedish drivers found on average they
sent one text message per month, with drivers between 18 and 24 sending three (Thulin
& Gustafsson, 2004), suggesting that younger drivers are much more likely to text whilst
driving than older drivers.
Note, while we may recognise it is important to separate texting from verbal
communication as the two tasks are very different, it is also appropriate to break down
texting into sending a text message and retrieving a text message, as these tasks differ
in some important characteristics (namely the greater complexity and duration of the
key presses required for sending a text). This distinction is addressed in this study.
Kircher, Vogel, Bolling, Nillson, Patten, Malmstrom, & Ceci, (2004) studied the effects of
receiving a text message on the performance of a small sample (ten) of experienced
drivers and found it significantly increased reaction times in a peripheral detection task
and generally reduced driver speed. From this we can infer that sending a message
would be even more detrimental to performance than receiving a message, however, as
Published Project Report
TRL
4
PPR 367
Kircher et al. did not examine the impact of sending messages, this assumption can not
be supported by their findings.
An attempt to measure the performance effects of retrieving and sending text messages
on young drivers was completed by Hosking, Young and Regan (2006) at Monash
University. They measured the effect of texting on young driver performance using the
advanced driving simulator located at Monash University Accident Research Centre
(MUARC). Twenty young adults (18-21) completed two drives which contained eight
“critical events” (e.g. avoiding a pedestrian, changing lane in accordance with traffic
signs, etc). During the drive participants were required to send and retrieve text
messages. Several measures of driving performance were found to be impaired during
the texting whilst driving condition. When texting, participants:
• spent 40% time looking away from the road environment compared to 10%
when undistracted
• were less consistent at maintaining appropriate positioning of their vehicle in
their lane (70% more variability in lateral lane positioning and 28% more lane
excursions).
• frequently failed to see signs instructing them to change lane (140% more
incorrect lane changes)
Participants were asked to complete a questionnaire post trial in which participants self-
rated any change in performance associated with driving whilst texting. These results
showed an awareness of a decline in performance, with 19/20 participants rating their
performance as worse when retrieving a message, and a full 20/20 participants rating
their performance as worse when sending a message.
Further support for drivers’ general awareness of the dangers of texting whilst driving
were also found by Ginsburg, Winston, Senserrick, Garcia-España, Kinsman, Quistberg,
Ross and Elliot (2008), who asked 5665 teenagers in the USA to rate various dangers
associated with driving. Texting whilst driving was rated as the second most likely
situation to make “a lot of difference” in driving safety. A modified and shortened list of
the risks used by Ginsburg, et al. was presented to participants in this trial.
Another key difference between using a mobile phone for verbal communication and
texting is that while adults generally share a similarly advanced degree of verbal skill,
the degree of experience/skill with texting can vary considerably between individuals. It
would seem reasonable to assume that while the performance impairments caused by
verbal communication are likely to be normally distributed, the performance impairments
of texting are likely to be skewed with older drivers showing greater reductions in
performance due to lack of experience with texting and diminished manual dexterity.
No research has been found to describe any performance differences resulting from
experience of texting, however, Chisholm, Caird and Lockhart (2007) examined a related
technology; the use of MP3 players whilst driving. This places similar demands on drivers
as texting whilst driving; both require a sequence of key inputs and are likely to divert
drivers’ vision away from the road environment and towards the device. Chisholm et al,
required 19 participants to operate an MP3 player whilst driving a simulator and
subjected them to a number of “critical events”. Their results demonstrated that
participants’ performance slowed responses to driving hazards while interacting with the
MP3 player declined somewhat, but a decrement still remained relative to the baseline
condition. This suggests that a driver’s risk of collisions whilst texting may decrease as
their experience/skill with texting increases.
Whilst research may have established that younger drivers are more likely to text, this
greater exposure to risks may be off-set (to a degree) by greater familiarity with texting
and increased confidence in their own abilities. While drivers are prone to overestimate
their own abilities, there is evidence that confidence is linked to driving performance.
Lesch and Hancock (2004) found that higher confidence amongst male drivers predicted
Published Project Report
TRL
5
PPR 367
increased driving performance. In contrast, higher confidence amongst female drivers
predicted lower driving performance. The reduction in performance amongst women was
accounted for by confidence rising with age; older women tended to be more confident
despite having slower reactions. However, confidence was a more accurate predictor of
performance amongst men regardless of age; when older male drivers reported more
confidence in their ability they did indeed exhibit superior performance regardless of
their age.
In order to determine whether confidence is predictive of performance when texting
whilst driving, this study required participants to complete a series of questions designed
to measure their confidence in their ability to text whilst driving and their general sense
of Self-efficacy (as measured by the IPIP NEO subscale). By comparing these outputs to
measures of experience with driving, and with general texting, the relationship between
confidence in general or confidence with texting and performance was investigated.
Further behavioural or attitudinal factors were explored. The RAC Foundation’s survey of
texting whilst driving (2008) described several personality archetypes that engage in this
behaviour. Analysis of these archetypes suggested that by examining participants
Immoderation, Liberalism and Self-efficacy (as measured by the IPIP NEO subscales),
we would have a description of their ability to resist using their phone to text whilst
driving, their degree of attachment to society’s rules and values, and how confident they
were regarding their own abilities in general. The predictive effects of these three factors
on texting whilst driving behaviour was investigated.
This study investigated two aspects of texting whilst driving using a high fidelity car
simulator:
• Driving performance. Quantitative measurements of driving performance were
recorded from participants as they drive a high fidelity car simulator when
engaged in a variety of text message tasks. This was to measure empirically the
changes in performance consequent from this behaviour. It was hypothesised
that:
o When writing/reading text messages, drivers would display:
Increased reaction times
Poorer car following ability
Poorer lateral lane control
Reduced speed
o Reductions in drivers’ performance would be greater when writing a text
message than when reading a text message.
• Subjective performance measures. Drivers reported how well they believe that
they drove in the simulator scenarios when engaged in text message tasks and
when undistracted. It was hypothesised that:
o Drivers shall report greater difficulty writing than reading text messages
Published Project Report
TRL
6
PPR 367
2 Method
2.1 Participants
Participants were recruited from the TRL participant database to take part in the study.
The aim was to recruit sixteen participants (eight male; eight female) aged between 18
and 25 years. Participants were selected if they described themselves as regular users of
text messaging to ensure that any performance effects seen are due to distraction from
the texting task and not due to drivers being unfamiliar with text messaging itself.
To ensure familiarity with phone operation, participants were required to use their own
mobile phones for the study. Participants had to use a phone that had a standard
alphanumeric keypad to ensure consistency in the task demand of message composition.
Participants whose personal phones had e.g. touch-screens; QWERTY keyboards for the
composition of text messages were excluded.
On arrival at TRL participants informed the trials manager of their mobile phone number,
the phone network that they use, and the phone make and model number (if known).
Participants were also asked to add four new contacts to their phone (all with the same
contact number) in order that they may send messages to these contacts over the
course of the drive.
Participants involved in the study were paid £35 as compensation for their time and
expenses incurred because of their participation.
2.2 Study design
In addition to their familiarisation drive, participants completed two test drives whilst at
TRL. A Texting drive in which they were required to perform a number of different text
messaging tasks (read; write; ignore) and a Control drive in which the participant drove
along the same route as the Texting drive but without having to perform the additional
texting tasks. To counterbalance any learning effect caused by completing the same
driving task twice, the order in which participants completed the Control and Texting
drives was alternated between participants.
2.3 Equipment
The TRL Driving Simulator (CarSim) consists of a medium sized family hatchback (Honda
Civic) surrounded by four 3 × 4 metre projection screens giving 210º front vision and
60º rear vision, enabling the normal use of the vehicle’s driving and wing mirrors. The
road images are generated by four PCs running SCANeR II software (manufactured by
Oktal) and are projected onto the screens by four Digital Light Processing (DLP)
projectors at a resolution of 1280 × 1024 pixels (giving a screen resolution of approx 13
pixels per inch). Images are refreshed at a rate of 60Hz (every 16.7msec) whilst data is
sampled at a rate of 20Hz (every 50msec). Electric motors supply motion with 3 degrees
of freedom (heave, pitch and roll) whilst engine noise, external road noise, and the
sounds of passing traffic are provided by a stereo sound system. Two studies
demonstrate the validity of the TRL simulator (Duncan, 1995; Sexton, 1997) and it can
be assumed that the current simulator system is at least as accurate as that used in the
Duncan and Sexton studies. Further details are provided in Appendix A.
An HP iPAQ 500 Series Voice Messenger mobile phone on the Vodafone network was
used by TRL to send /receive the text messages for the study (participants used their
own mobile phone/network provider).
Published Project Report
TRL
7
PPR 367
2.4 Familiarisation
Participants were required to complete a familiarisation drive prior to completing any of
the test drives in order to be comfortable with controlling the simulator vehicle and
driving in the virtual environment. This familiarisation drive was in a benign motorway
environment and lasted approximately ten minutes. It included a car following task in
which participants were required to drive at a safe and constant distance behind a lead
vehicle. During this task, white chevrons were included on the motorway (as used on
some sections of UK motorways) helping the participant to judge a safe distance to the
lead vehicle.
Figure 2.1 Car-following route with motorway chevrons
2.5 Participant instructions
Participants were given specific instructions before driving relating to the simulator task.
These are shown in the following text box:
Published Project Report
TRL
8
PPR 367
Participant instructions
Beside the car:
Please adjust the seat position and secure the safety belt. The car controls work in the
same manner as any normal car and it operates with a manual gearbox.
You need to make sure the car is in neutral when you start it and it needs plenty of revs,
otherwise it has a tendency to stall.
It is important that you drive as you would normally. We don’t want you to drive as if
you are on a driving test nor as if the simulation is a computer game. We are not here to
judge your driving, so please do not feel anxious.
A red bar like the one you can see on the screen now will appear at some point during
your drive. There is also a buzzing noise that will sound at some point during the drive.
When you hear the buzzing noise, or see the red bar please press the clutch pedal as
quickly as you possibly can. .You will hear the buzzing sound about 20 seconds into this
drive as a practice to help you recognise it.
The drive will start on the motorway with normal traffic. After some while, the motorway
will end and you will reach a series of bends. You should try to keep to 40 mph through
this section and the simulator will assess your ability to keep to the centre of your lane
through the bends.
After the series of bends, you will drive on the motorway again. After a period of time
you will see a vehicle in front of you. Please pull up behind this vehicle and follow it,
doing your best to keep at a safe and constant distance behind it. A voice instruction will
let you know when the car following task has finished.
There will be voice instructions to remind you about these tasks.
Text drive only
During this drive you will be asked to send and receive text messages, please only read
text messages when you are asked to. Please text as you normally would i.e. using short
words and predictive text.
From the control room:
Start: Please start the engine using the ignition key and proceed, driving as you would
normally
2.6 Questionnaires
Over the course of their involvement in the study, participants completed a
questionnaire covering a range of items:
• Demographic information
• Participant experience of simulator sickness over the course of the trial
• Questions about the received text messages (to confirm that participants read the
text)
• Participants’ self assessment of performance in the simulator drives
• Participants’ self assessment of the effect of texting on their driving performance
• Participants’ assessment of relative risk in a range of risky scenarios
• Participants’ perception of the legality of reading/writing text messages whilst
driving
• Participants’ reported mobile phone usage
Published Project Report
TRL
9
PPR 367
The questionnaire included the Driver Behaviour Questionnaire (DBQ; Reason, Manstead,
Stradling, Baxter, & Campbell, 1990). The DBQ gives a relative score in three
dimensions of driving behaviour: Violations, Errors, and Lapses. The correlations
between scores on these dimensions and the driving behaviour measures recorded in the
simulator were measured.
Extracts of the International Personality Item Pool (IPIP; Goldberg, Johnson, Eber,
Hogan, Ashton, Cloninger & Gough, 2006) were also used in an attempt to describe the
differences between personality types and their frequency of texting: in effect, why
participants choose to text. Three facets were chosen; these are shown in Table 2.1.
Table 2.1 IPIP facets used and their hypothesised relation to texting whilst
driving
IPIP facet Supposed relation to texting whilst driving
Self-efficacy An individual’s confidence in their ability to drive and text
Openness to values An individual’s tendency to follow the rules (in this case, of the road)
Impulsiveness An individual’s ability to resist temptation to text
2.7 Route design
The simulator route driven by participants consisted of four sections with smooth
naturalistic transitions between each section.
Table 2.2 The road sections used for the simulator trial
Section
Description Length Configuration
1 Motorway 1 28.1km 3 lane motorway plus hard shoulder in each direction.
Light traffic present
2 Two loops 7.3km Each loop is a two-lane ‘figure 8’ with a long left turn
and long right turn separated by a short straight
3 Car following 13.0km 3 lane motorway plus hard shoulder in each direction.
One vehicle present that the participant is required to
follow at a steady distance
4 Motorway 2 11.6km 3 lane motorway plus hard shoulder in each direction.
Light traffic present
Total
60.0km
In the loops section, participants were instructed to attempt to stay in the centre of their
lane and to drive at 40mph
1
.
In the car following section, participants were instructed to follow the lead vehicle at safe
and steady distance (as they would have experienced in the familiarisation drive). The
lead vehicle continuously varied its speed sinusoidally between 70km/h (43.8mph) and
110km/h (68.8mph) over a period of 20 seconds.
1
In previous studies at TRL that have used a similar route design, participants were required to drive through
the loops section at 60mph. However, in pilot trials, it was found that combining the text messaging tasks with
safe route navigation at this speed was too difficult. Consequently, a lower speed was chosen to reduce the risk
that participants would lose control of the vehicle, which would have negative implications both for the
participant’s experience of the trial and data collection.
Published Project Report
TRL
10
PPR 367
2.7.1 Reaction time events
Over the course of both the Texting and Control drive, participants were required to
respond to trigger stimuli on four occasions. The response was to depress the clutch
pedal as quickly as possible. Clutch depression is measured from 0 (foot off clutch) to 1
(clutch fully depressed). The threshold for clutch activation was 0.1 (10% clutch
depression). If clutch depression was greater than 10% at the time of the reaction time
trigger, the event would have been ignored but this did not occur in any of the trials. If
participants failed to respond within 10 seconds, this was treated as a missed event.
In three of the reaction events, the trigger stimulus was a short auditory tone (60dB;
0.45 seconds duration; 333Hz). The fourth reaction time trigger event was the
presentation of a red bar stimulus above the carriageway and ahead of the driven
vehicle across all motorway lanes. This is shown in Figure 2.2.
Figure 2.2 The trigger stimulus for the visual reaction time event
The reaction time tasks were triggered through the drive at times when the participant
was likely to be engaged in a texting task in the Texting drive. The reaction time triggers
were in exactly the same place for the Control drive.
2.7.2 Texting tasks
Through the course of the drive participants were required to perform a succession of
different text messaging tasks.
Published Project Report
TRL
11
PPR 367
2.7.2.1 Writing text messages
Participants had to write five text messages in the drive. All required an approximately
similar number of characters with some possibilities to apply SMS language if desired.
The instructions as to what to write and the message recipient were delivered as
automated verbal messages in the simulation. As shown in section 2.5, participants were
instructed to compose the text in their own usual style. This included using predictive
text and applying SMS language. The first message was included as practice to ensure
participants were comfortable what was required of them.
Table 2.3 shows the text messages that participants were required to compose.
Table 2.3 Text messages composed by participants whilst driving
Message Recipient Section Message Characters
Practice Adam 1 “I am driving a great car simulator” 34
1 Adam 1 “Happy birthday Have fun at the party” 36
2 Brian 2 “Nice to see you at the cafe yesterday” 37
3 Claire 3 “Dont worry Have a nice time in Paris” 36
4 Dawn 4 “Sorry about your ankle Get well soon” 36
2.7.2.2 Reading text messages
Participants were sent two text messages over the course of their drive. Before receiving
messages, participants were informed by an automated voice instruction that they were
about to receive a message and that they would need to read the message in order to be
able to answer the questionnaire at the end of the drive. Table 2.4 shows the messages
that participants received.
Table 2.4 Text messages received by participants whilst driving
Message Section Message
1 1 “Edward has forgotten his BOWTIE for the wedding”
2 2 “Fiona won the SILVER medal in the 100m sprint”
2.7.2.3 Ignored text message
In addition to writing and reading text messages, participants were also sent a text
message without any forewarning. As shown in section 2.5, participants were informed
only to read messages that they had been instructed to read. So this event represented
the distraction caused a message that participants knew they had received but were
unable to read. The message is shown in Table 2.1.
Table 2.5 Text message to be ignored by participants whilst driving
Message Section Message
1 3 “Text message from TRL – please ignore”
Published Project Report
TRL
12
PPR 367
2.7.2.4 Control messages
To compare the difficulty of texting whilst driving against composing text messages with
no other distractions, participants were timed composing some comparable text
messages. These were as follows:
Table 2.6 Timed text messages composed by participants without distraction
Recipient Message Characters
Adam “Best of luck for your driving test today” 40
Brian “Well done Looking forward to the wedding” 40
Claire “Please can you bring red wine tonight” 37
Dawn “Where did you get those new trousers” 36
Published Project Report
TRL
13
PPR 367
2.7.3 Overall design
Figure 2.3 shows how the various texting tasks and reaction time events were initiated
over the course of the drive. Note that the Control drive was the same as the Texting
drive with the exception that participant was not required to interact with their phone.
Motorway 1 Curve following Car following Motorway 2
Write SMS 1
Write SMS 2
Write SMS 3
Read SMS 1 Auditory RT
Auditory RT
Trial start
Trial end
Auditory RT
Write SMS 4 Visual RT
Ignored SMS
SMS task RT task
Read SMS 2
Write SMS p
Motorway 1 Curve following Car following Motorway 2
Write SMS 1
Write SMS 2
Write SMS 3
Read SMS 1 Auditory RT
Auditory RT
Trial start
Trial end
Auditory RT
Write SMS 4 Visual RT
Ignored SMS
SMS task RT task
Read SMS 2
Write SMS p
Figure 2.3 Overall design of Texting drive
Published Project Report
TRL
14
PPR 367
2.8 Trial procedure
The trial proceeded as shown in Table 2.7.
Table 2.7 Trial procedure
Time from start Activity Duration
0 Welcome and introduction 5
5 Simulator: Baseline drive 10
15 Drive 1 (Texting or Control) 40
55 Control messages 40
95 Drive 2 (Control or Texting) 40
135 Questionnaire 30
165 Depart
Published Project Report
TRL
15
PPR 367
2.9 Recorded simulator data
The simulator recorded the following data about participants’ simulator driving at 20Hz
through each drive.
Table 2.8 Data recorded by the simulator at 20Hz
Data Notes
Time Time elapsed since the start of the trial
X position of
interactive vehicle
The X position of the interactive vehicle within the map of the
simulated environment.
Y position of
interactive vehicle
The Y position of the interactive vehicle within the map of the
simulated environment.
Z position of
interactive vehicle
The Z position of the interactive vehicle within the map of the
simulated environment.
Speed Current speed of the interactive vehicle
Distance through trial Distance travelled by participant relative to the start of the virtual
road
Lateral distance from
centre of road
The distance of the centre of the interactive vehicle from the centre of
the road
Headway The distance headway between the interactive vehicle and the back of
any vehicle ahead.
Time Headway The time headway between the interactive vehicle and the back of any
vehicle ahead.
Accelerator pedal Current proportion of accelerator pedal depression.
Brake pedal Current proportion of brake pedal depression.
Clutch pedal Current proportion of clutch pedal depression.
Steering wheel Current angle of steering wheel rotation
2.10 Calculation
Manipulation of the simulator data was conducted using Microsoft Excel 2002. Statistical
analyses were conducted using SPSS 14.0. In statistical tests, p values of less than 0.05
were taken to be significant. Where shown, error bars indicate the 95% confidence
interval on the mean.
Published Project Report
TRL
16
PPR 367
3 Results
3.1 Participants
Seventeen participants (selected from the TRL participant database) took part in the
study. This consisted of nine females and eight males. As it was expected that young
drivers are more likely to be habitual text message users, the age range of participants
identified for this study was chosen in light of this, with all participants aged between 17
and 24 (M = 20.4, SD = 1.84). Only participants who described themselves as regular
users of text messaging were selected. Furthermore, it was required that all participants
owned a mobile phone with a standard alphanumeric keypad. The phone makes and
network providers used by participants are described in Appendix B.
3.2 Reaction time (RT) tasks
In the simulator drives, participants were required to respond to three auditory tones
and one red bar visual stimulus by pressing the clutch as quickly as possible.
- Auditory RT task 1 coincided with Read text message 1 in section 1
- Auditory RT task 2 coincided with Write text message 2 in section 2
- Auditory RT task 3 coincided with Write text message 3 in section 3
- Visual RT task coincided with Write text message 4 in section 4
3.2.1 Response rate
In each of the Control and Testing drives there should have been 68 responses (17
participants × 4 RT events per drive). Participants failed to respond on 6/68 occasions in
the Control drive. Participants failed to respond on 14/68 occasions in the Texting drive.
These proportions near significance in a chi-square comparison (χ
2
(1) = 3.752; p =
0.053) suggesting that the texting tasks interfered with participants’ ability to respond to
the reaction time tasks.
3.2.2 Reaction times
In the reaction time tasks where participants successfully responded to the stimulus,
Figure 3.1 shows the mean reaction time to each of the stimuli. Note that Auditory 1 was
triggered during a read text message task whilst the other RT tasks were triggered
during a write text message task.
Published Project Report
TRL
17
PPR 367
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6
Auditory 1 Auditory 2 Auditory 3 Visual
RT task
RT (sec)
Control
Texting
Figure 3.1 Mean reaction times to each of the RT tasks
Figure 3.1 shows that in each of the RT tasks, the mean RT was greater in the Texting
drive than in the Control drive. Paired sample t-test comparisons show that this
difference was significant in each case apart from Auditory 3 (Auditory 1: t(13) = -
5.689, p < 0.001; Auditory 2: t(13) = -2.904; p = 0.012; Auditory 3: t(12) = -0.654; p
= 0.526; Visual: t(12) = -2.442; p = 0.031).
3.3 Analyses by section
In previous similar studies conducted by TRL (e.g. Burns, Parkes, Burton, Smith & Burch,
2002; Parkes, Luke, Burns & Lansdown 2007), comparisons between control and
distracted conditions have been conducted over complete sections of the drive; for
example, comparing performance through the entire loop sections. This is less
appropriate for this study since the texting tasks are not being performed continuously
through each section.
Some basic analyses were performed to compare behaviours in sections and for each
complete drive to investigate whether the texting drives may have had significant and/or
long lasting effects on behaviour.
3.3.1 Speed
Figure 3.2 shows the maximum observed speed across participants over the course of
the trial.
Published Project Report
TRL
18
PPR 367
72
74
76
78
80
82
84
86
88
90
Control Texting
Drive
Mean maximum speed (mph)
Figure 3.2 The mean of Maximum speed values observed across participants in
each drive
The maximum speed observed across participants was significantly lower in the Texting
drive (t(16) = 2.297; p = 0.035). A similar pattern is seen for the maximum speed
observed in section 1 (Control: 82.4mph vs. Texting: 78.9mph; t(16) = 2.206; p =
0.042).
Figure 3.3 shows the mean speed observed within section 3 (Car following) of the drive.
Note that due to the requirements of the car following task within section 3, there was
much less variability in the observed speeds.
Published Project Report
TRL
19
PPR 367
53.6
53.7
53.8
53.9
54.0
54.1
54.2
54.3
54.4
Control Texting
Drive
Mean speed (mph)
Figure 3.3 The mean speed values observed across participants in Section 3 of
each drive
Again, significantly lower speeds are observed in the Texting condition (Control:
54.13mph vs. Texting: 53.97mph; t(16) = 2.216; p = 0.042).
The overall pattern of results regarding speed is indicative that participants felt less
comfortable when driving and having to complete the texting tasks and chose to reduce
their speed in order to manage their perceived level of risk to a subjectively acceptable
level.
3.3.2 Variation in lane position
Variability in lane position is a commonly used measure of driving performance and is
usually measured as the standard deviation of lane position (SDLP). Additional task
demands and/or sub-optimal driver physical state are reflected in increased swerving
behaviour (e.g. Brookhuis, de Vries & de Waard, 1991; Burns et al., 2002). SDLP was
available for measurement in this trial since position across the road was measured in
metres throughout the recorded drives. Figure 3.4 shows SDLP within section 3.
Published Project Report
TRL
20
PPR 367
0.60
0.62
0.64
0.66
0.68
0.70
0.72
0.74
0.76
0.78
0.80
Control Texting
Drive
Mean SDLP
Figure 3.4 Mean SDLP across participants observed in section 3.
The difference shown in Figure 3.4 indicates significantly greater variability in lane
position when texting (t(16) = -2.175; p = 0.045). Note that this in spite of participants’
significant reduction in speed within section 3. The comparison of overall SDLP within
section 2 (Loops) also nears significance (t(16) = -2.077; p = 0.054).
3.4 Analyses by texting episode
An intuitively more appealing approach was to examine the periods during which the
participant was engaged in a texting task and to compare behaviour in the identical
section of the Control drive. This analysis region was determined in two ways. For the
write text message tasks, the start point was taken as the position of the driven vehicle
at the end of the voice instruction relating to that message. The endpoint was taken as
the position of the driven vehicle when the participant sent the composed message. For
the read and ignore text message tasks, the start point was taken as the position of the
driven vehicle when the text message arrived at their phone. There was no easily
discernible endpoint to use for these tasks so the position of the vehicle 60 seconds after
the start point demarcated the analysis region. By comparing behaviour in the analysis
region when texting to the equivalent analysis region in the control drive, a more precise
evaluation of the direct impairment caused by the texting task may be obtained.
There are two key assumptions in this approach. Firstly, that the distraction caused by
the texting task is immediate (whether that means they begin writing the message
directly after the end of the voice instruction, that they try to read the message
immediately after it arrives, or they are distracted by the arrival of a message that they
must ignore). Secondly, that the impairment caused by the text message task is
confined to the analysis region. This is by no means certain. Redelmeier & Tibshirani
(1997) found that accident risk increased significantly not only during a mobile phone
conversation but also in the following minutes. Consequently, defining such endpoints to
the analysis means that some driving impairment may not be accounted for in the
analysis.
For each text message task, four key measures were taken:
Published Project Report
TRL
21
PPR 367
• Mean speed
• Standard deviation of speed
• SDLP
• Maximum observed speed
For Sections 2 (Loops) and 3 (Car following), some additional measures were taken
based on the additional task demands in these sections.
3.4.1 Section 1 (Motorway 1)
After the initial (practice) write text message task, there were two analysed text
message tasks within section 1; a write message task (Write 1) and a read message task
(Read 1). The Read 1 task was associated with an auditory RT event.
3.4.1.1 Read 1
Comparisons of behaviour in Read 1 revealed very few differences in the measures
taken. The only comparison that neared significance was that between the maximum
speed observed within the Read 1 analysis region (Control: 77.1mph vs. Texting:
73.1mph; t(16) = 2.216; p = 0.042). This fits with the observation that overall speeds
were lower in Section 1.
3.4.1.2 Write 1
As with Read 1, the measures taken did not reveal large differences in behaviour in the
Write 1 task. None of the paired comparison t-tests neared significance (p > 0.15 in
each case).
3.4.2 Section 2 (Loops)
In Section 2, participants completed two repeats of a figure-eight with variable radius
loops. Within the first loop, participants were required to complete a write message task
(Write 2) and a read message task (Read 2).
3.4.2.1 Write 2
The key difference between the Control and Texting drives was the increase in SDLP
when composing the message as illustrated by Figure 3.5. The paired comparisons t-test
produced a highly significant result (t(14) = -3.137; p = 0.00728). Other comparisons
failed to reach significance (p > 0.14 in each case).
Published Project Report
TRL
22
PPR 367
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
Control Texting
Drive
Mean SDLP
Figure 3.5 Mean SDLP in the Write 2 task relative to that observed in the
Control drive
This big difference in lateral position variability warranted further investigation. The
number of occasions in which an edge of the driven vehicle departed from the driven
lane was counted for each participant. Figure 3.6 shows the frequency count of lane
departures for all participants in the Write 2 task compared to the Control drive.
42 departures
(by 9/15 participants)
4 departures
(by 3/15 participants)
0
10
20
30
40
50
Control Texting
Drive
Total number of lane departures
Figure 3.6 Total number of lane departures observed in the Write 2 task
compared to those in the Control drive
Published Project Report
TRL
23
PPR 367
Figure 3.5 gives a clear demonstration of the impairment to vehicle control caused by
attempting to compose a text message whilst driving through a series of bends. More
lane departures are observed by more participants when engaged in the text messaging
task.
3.4.2.2 Read 2
As with Write 2, the clearest difference between the Control and Texting drives was in
SDLP with the texting task causing SDLP to increase. However, the paired comparison t-
test of SDLP failed to reach significance for this task (t(16) = -1.681; p = 0.112). All
other tests failed to reach significance. As before, however, the total number of lane
departures was calculated for the Read 2 task and is shown in Figure 3.7.
18 departures
(by 4/17 participants)
8 departures
(by 3/17 participants)
0
5
10
15
20
25
Control Texting
Drive
Total number of lane departures
Figure 3.7 Total number of lane departures observed in the Read 2 task
compared to those in the Control drive
Again, it can be seen that lane departures are more frequent and made by more
participants when they were completing the read text task, although the differences are
less clear cut than in the Write 2 comparison.
3.4.3 Section 3 (Car following)
In Section 3, participants were required to follow a lead vehicle at a subjectively safe,
constant distance whilst the lead vehicle varied its speed sinusoidally. Whilst engaged in
the car-following task, participants were required to write a text message (Write 3) and
to ignore an unexpected incoming message (Ignore).
3.4.3.1 Write 3
Paired-comparisons t-tests across all the main measures reached (or neared)
significance in the Write 3 task. When texting, mean speed was lower (t(16) = 2.213; p
= 0.0418), maximum speed was lower (t(16) = 4.067; p < 0.001), speed variability was
lower (t(16) = 3.974; p = 0.00109), and SDLP was higher (t(16) = -2.06; p = 0.0555).
Published Project Report
TRL
24
PPR 367
Figure 3.8 shows the differences in mean and maximum speed for the Write 3 task.
50
52
54
56
58
60
62
64
66
68
70
Write 3 mean speed (mph) Write 3 max speed (mph)
Texting task
Speed (mph)
Control
Texting
Figure 3.8 Mean speed and Maximum speed observed in the Write 3 task
relative to that observed in the Control drive
It may seem counterintuitive that participants should reduce speed variability when
composing the text message in the Write 3 task. However, remember that the task in
Section 3 was to remain at a constant distance behind a lead vehicle that varied its
speed. Therefore, the participant should vary their speed in order to maintain this
constant distance. That reduced variability is observed when texting is indicative of
performance impairment in the car following task. This can be investigated by examining
the time headway relative to the lead vehicle during the texting task. Figure 3.9 shows
the mean, standard deviation, and minimum time headways observed for the Write 3
task in the Texting and Control drives.
Published Project Report
TRL
25
PPR 367
0
1
2
3
4
5
6
7
8
Write 3 mean time
headway (sec) Write 3 SD time headway Write 3 minimum time
headway (sec)
Texting task
Time headway
Control
Texting
Figure 3.9 Mean, Standard Deviation of, and Minimum time headway relative to
the lead vehicle for the Write 3 task during the car following section relative to
that observed in the Control drive
It can be seen that participants maintained higher mean and minimum time headways to
the lead vehicle in the Texting drive. This may have been a risk mitigation tactic in the
awareness that they had poorer control of their vehicle. However, time headway
variability is much lower in the Control drive due to participants maintaining a better
constant distance behind the lead vehicle. Paired comparison t-tests are significant (or
near significant) for all three time headway comparisons (Mean time headway: t(16) = -
4.344; p < 0.001. SD time headway: t(16) = -2.043; p. = 0.0579. Minimum time
headway: t(14) = -3.346; p = 0.00480).
As stated above, the comparison of SDLP across drives neared significance. This is
shown in Figure 3.10.
Published Project Report
TRL
26
PPR 367
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Control Texting
Drive
Mean SDLP
Figure 3.10 Mean SDLP in the Write 3 task relative to that observed in the
Control drive
Figure 3.10 demonstrates that although participants may be able to manage their risk of
collision with a vehicle in front by increasing headway distances, they are unable to
compensate for the impairment in lateral control. As with Write 2, the variation in lane
position may have lead to an increase in the number of lane departures observed in the
Write 3 task. The frequency count is shown in Figure 3.11.
Published Project Report
TRL
27
PPR 367
18 departures
(by 4/17 participants)
0 departures
(by 0/17 participants)
0
5
10
15
20
25
Control Texting
Drive
Total number of lane departures
Figure 3.11 Total number of lane departures observed in the Write 3 task
compared to those in the Control drive
As with the Write 2 task, Figure 3.11 shows a stark contrast between the Texting and
Control drives. The increase in SDLP observed in the Texting condition led to occasions
when the edge of the driven vehicle drifted into the adjacent lane in a task where the
participant was required to remain behind a lead vehicle travelling in one lane only.
3.4.3.2 Ignored
The ignored message did not appear to cause any changes in participants’ measured
driving behaviour. The observed mean values for the four key measures were relatively
close and paired samples t-tests did not near significance for any comparison (p > 0.23
in each case).
3.4.4 Section 4 (Motorway 2)
In Section 4, participants returned to the standard 3-lane motorway environment with
moderate traffic levels. Within section 4, participants were required to write a text
message (Write 4) and the visual RT stimulus was displayed concurrently.
3.4.4.1 Write 4
In the Write 4 task, only one key measure reached significance in the paired
comparisons t-tests; that comparing mean speed between the Texting and Control drive
(t(15) = 3.073; p = 0.00773). Once again, participants chose to adopt a significantly
lower speed when texting (Control: 73.1mph; Texting 681.mph). The comparisons of
maximum speed (lower in Texting drive) and speed variability (higher in Texting drive)
also neared significance (p = 0.0619; p = 0.099 respectively.
Published Project Report
TRL
28
PPR 367
3.4.5 Texting completion times
The time taken to complete composition of each text message was recorded. The mean
times taken for each message are shown in Figure 3.12.
0
10
20
30
40
50
60
70
80
Write 1 duration Write 2 duration Write 3 duration Write 4 duration
Text message task
Text completion time (sec)
Figure 3.12 Text completion times when driving
Repeated measures t-tests between each message revealed that Write 1 was completed
significantly more quickly than Write 2 (t(14) = -5.199; p < 0.001) and than Write 4
(t(15) = -2.275; p = 0.038). Write 3 was completed significantly more quickly than
Write 2 (t(14) = 3.496; p = 0.004). The increased task difficulty in the loops section is
likely to account for the increase in completion time for Write 2. The mean completion
time overall was 63.7 seconds.
3.5 Gender differences
Approximately equal numbers of male and female participants were recruited for the
study, providing an opportunity to investigate the relative affect of the text messaging
tasks on male and female drivers by repeated measures ANOVA tests on the recorded
simulator measures across the factor of gender. Significant interaction results are
described below.
• SDLP in Write 2 (F(1, 13) = 5.774; p = 0.032)
• Mean speed in Write 3 (F (1, 15) = 4.920; p = 0.042)
Plots for the interactions are shown in figure
Published Project Report
TRL
29
PPR 367
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Male Female
Write 2 SDLP
Mean SDLP
Control
Texting
52.0
52.5
53.0
53.5
54.0
54.5
55.0
55.5
56.0
56.5
Male Female
Write 3 Mean speed
Mean speed (mph)
Control
Texting
(a) (b)
Figure 3.13 Plots of the interaction between gender and (a) SDLP in Write 2
and (b) Mean speed in Write 3
Male participants appear to show less impairment in SDLP when texting in the loops
section. Female drivers appeared to be more aware of the impairment as they showed
greater reductions in speed in the Texting drive. By contrast, male drivers showed a
negligible reduction in speed. Note that similar ANOVA tests on text completion times
showed that there was no significant differences between male and female participants
in the time taken to complete the text messages composed whilst driving (p > 0.5) or
those composed in controlled conditions (p > 0.3).
This reduction in SDLP impairment in Write 2 for male drivers resulted in fewer lane
departures. Of the 42 lane departures observed in Write 2 in the Texting drive, 39 were
by the female drivers. In the Write 2 analysis region of the Control drive both males and
females made three lane departures. A chi-square test comparing the frequency of lane
departures by male and female drivers for the Write 3 analysis region in the Control and
Texting drives reveals that this difference is significant (χ
2
(1) = 8.816; p = 0.003).
Furthermore, all eighteen observed lane departures in the Write 3 analysis region were
by female drivers. However, when the frequency count of the 17 lane departures
observed when reading text messages in the Read 2 analysis region is examined, the
distribution is more even (7 male; 10 female).
3.6 Error rate
The messages sent by participants within the Texting drive were recorded for analysis.
There was variation in the style of composition (use of SMS language, punctuation,
capitalisation, spelling) but, given the dual task demand, accuracy was remarkably high.
Two participants failed to complete messages (one participant missed one of the four
test messages, the other missed two of the four test messages). Of the remaining 65
messages, only two had significant errors that might compromise understanding by the
recipient:
• Participant 10
o “Niåe to se u at the cafe y'day”
Should be “Nice to see you at the café yesterday”
o “Sorry abovt ur ankle get well soon.”
Should be “Sorry about your ankle. Get well soon”
• Participant 17
o “Sorry about youre ankle gdu well soon”
Should be “Sorry about your ankle. Get well soon”
Published Project Report
TRL
30
PPR 367
A table of the text of all composed messages is shown in Appendix B.
3.7 Patterns of mobile phone use
As a part of the post-trial questionnaires, participants were required to describe their
habitual patterns of mobile phone use. These questions were asked to ensure that all
participants were of a similar level of experience/familiarity with mobile phones, and
texting in particular.
3.7.1 Familiarity with mobile phones and text messaging
Participants reported owning a mobile phone for a mean duration of 7.5 years (range =
5-12, SD = 2.06). This duration of ownership suggests that participants were familiar
with the normal operation of mobile phones for making calls and sending text messages.
The frequency with which participants used their mobile phones for verbal
communication and text messaging was obtained. As can be seen in Figure 3.14, the
frequency of verbal communication was evenly spread between the first four categories,
with no participants reporting using their phone more then 30 times a week. In contrast,
reported use of text messaging was heavily skewed towards the “21 to 30” category,
with 11 out of 17 participants selecting this category. This suggests texting is very
popular amongst our participants and their exposure to incidents of text messaging
(either composing or receiving) while driving is likely to be significantly higher than their
exposure to incidents of verbal communication. Therefore, texting whilst driving might
be the greatest source of risk associated with mobile phone use whilst driving.
0
2
4
6
8
10
12
1 to 5 6 to 10 11 to 20 21 to 30 30+
Number of uses
Number of participants
Spoken
Text
Figure 3.14 Weekly use of mobile phone for spoken conversations and text
messaging.
A range of makes and models of mobile phones were used by the participants: 8 Nokia,
4 Samsung, 4 Sony Ericcson, 1 Sagem. Furthermore, 8 of the participants used
predictive texting and 9 did not. Participants were also asked to rate how easy they felt
their phone was to operate. As can be seen in Figure 3.15, participants mostly rated
their phones as easy to used (M = 76.01, SD = 29.70). Note the presence of an outlier
was identified in the data. One participant rated their phone as much more difficult to
use than the others (specifically, its ease of use was rated as 10 out of 100). This may
be because the participant genuinely finds their phone difficult to use, or perhaps the
participant might have misread the scale leading to a reversed score.
Published Project Report
TRL
31
PPR 367
Figure 3.15 Participants’ ratings of ease of use of their mobile phones
3.7.2 Storage and method of use of mobile phones whilst driving
Several questions regarding how participants store their mobile phone when driving.
Specifically, participants were asked:
• If they bring their phones with them when driving
• If they leave their phones switched on when driving
• If they leave them on silent
• If they use a phone cradle
• If they use their phone ‘hands-free’
For three of the questions all of the participants chose the same answer. All seventeen
participants reported: taking their mobile phones while them “always” when driving,
“always” leaving their phones switched on when driving, and none of the participants
reported using a cradle for their mobile phone. These results indicate that all of the
participants may be at risk of non-compliance with the law, as it is illegal to use a mobile
phone when in a car if you are required to touch your phone to do so (unless it is in a
cradle).
There was more variation in the participant’s tendency to leave their phones on silent
and whether they used their phones ‘hands-free’. In Figure 3.16, we can see that only
three participants always switch their phone on silent when driving. In contrast, a
combined number of 13 participants reported “sometimes”, “occasionally” or “never”
doing so. This suggests that the majority participants may be aware of their phone
receiving an incoming call or message while they drive (and are exposing themselves to
the temptation of answering it).
Published Project Report
TRL
32
PPR 367
Never (6)
Occasionally (1)
Always (3)
Often (1)
Sometimes (6)
Figure 3.16 Frequency with which participants leave their mobile phones on
silent
In Figure 3.17, we can see that only two participants report using their phone hands-
free, with another three reporting using it “sometimes”. Twelve participants reported not
using their mobile hands-free at all.
Do not use (12)
Do use (2)
Sometimes use (3)
Figure 3.17 Frequency with which participants use their mobile phones 'hands-
free'
Together these results show that the majority of participants are likely to leave their
phones switched on and in non-silent mode when they drive. They also do not use a
cradle for their phones, nor do the majority of them use their phones ‘hands-free’.
3.7.3 Baseline texting completion time
Before the trial, participants were required to complete four baseline texting tasks to
measure their speed of texting when undistracted. Fifteen of the participants completed
these baseline texts and a box plot was produced from these results (see Figure 3.18).
As can be seen in Figure 3.18, the mean time per text was 22.66 seconds (Min = 10.75;
Max = 42.75; SD = 8.51). The degree of variation in texting speed was within
acceptable bounds, indicating participants were reasonably similar in their texting speed.
Published Project Report
TRL
33
PPR 367
Baseline text completion
Figure 3.18 Distribution of participants’ mean baseline texting completion times
3.7.4 Comparison of text completion time when driving vs. baseline
Figure 3.19 shows the difference between the time taken to complete text messages
when driving in comparison to the time taken to complete similar text messages in the
undistracted, baseline condition.
0
10
20
30
40
50
60
70
80
Driving Baseline
Text messaging situation
Mean completion time (sec)
Figure 3.19 Text completion times when driving compared to the baseline
condition
Figure 3.19 shows that the text messages took considerably longer to complete when
driving. This is unsurprising given that attention must be split between texting and
driving in the dual task condition. The difference between the average completion times
is highly significant in a t-test (t(123) = 16.84, p < 0.001).
Published Project Report
TRL
34
PPR 367
3.8 Perceptions of the legality of mobile phone use while driving
Participants’ beliefs about the legality of texting whilst driving were garnered after the
trials. The questions were split into two sections: participants understanding of the laws
as they now stand, and their beliefs about whether these behaviours should be illegal. In
both sections participants were asked four questions (to which they could answer “legal”,
“illegal” or “not sure”):
• Is it currently legal [or should it be legal] to use your phone whilst driving to…
o …send a text message if it is in a cradle?
o …read a text message if it is in a cradle?
o …send a text message if you are using it handheld?
o …read a text message if you are using it handheld?
In Figure 3.20, we can see a series of pie charts describing that the majority of
participants believe it is illegal to text whilst driving in all circumstances. The first two pie
charts in Figure 3.20 demonstrate participants’ view of the legality of sending and
reading a text when the phone is in a cradle. Eleven participants thought it was illegal to
send a text message whilst driving and nine thought it was illegal to read a message.
These participants were incorrect. Only two correctly responded that it was legal to send
a text message as long as the phone is in a cradle, and four participants responded that
it is legal to read a message. In fact, current legislation does not explicitly prohibit
texting whilst driving provided the phone is secured in a cradle. However the UK
Highway Code (2007) does state that:
“You MUST exercise proper control of your vehicle at all times. You MUST NOT use a
hand-held mobile phone, or similar device, when driving or when supervising a learner
driver, except to call 999 or 112 in a genuine emergency when it is unsafe or impractical
to stop. Never use a hand-held microphone when driving. Using hands free equipment is
also likely to distract your attention from the road. It is far safer not to use any
telephone while you are driving - find a safe place to stop first.”
(Laws RTA 1988 sects 2 & 3 & CUR regs 104 & 110; Highway Code, The Stationery
Office, 2007)
The third and fourth pie charts show participants’ responses in relation to texting using a
handheld phone. We can see that the response pattern was almost identical to the first
two questions. Specifically, this time eleven participants thought that it was illegal to
text using a handheld phone to send a message, and nine thought it was illegal to read a
text. However, this time, they were correct.
Figure 3.20 Participants’ understanding of the current legality of mobile phone
use
After asking participants for their beliefs about the current state of the law they were
asked whether they thought the same four activities should be illegal (or not). In Figure
3.21, a series of pie charts describes their responses, and from them we can see a broad
Legal
Illegal
Not sure
Missing
Send message
Phone in cradle
Read message
Phone in cradle
Send message
Phone handheld
Read message
Phone handheld
2
4 1
2
1
11
3
1
1
9
3
11
3
3
9
4
Published Project Report
TRL
35
PPR 367
consensual trend towards the attitude that sending a message should be illegal,
regardless of whether the phone is in a cradle or handheld (only one participant felt it
should be legal in either circumstance). Furthermore, there was a good deal of
agreement that reading a text message should be illegal. However, a few participants
felt reading messages was acceptable (Four for when the phone is in a cradle and three
for when it is handheld).
Figure 3.21 Participants’ beliefs about whether various behaviours should be
legal
3.9 Perceptions of the relative risks of driving behaviours
Participants were asked to judge the degree of risk presented by a variety of driving
behaviours (14 in total). The scale stretched from “Less dangerous” (0%) to “More
dangerous” (100%). The results of this section of the questionnaire can be seen in Table
3.1. Participants reported alcohol, racing and fatigue to be three greatest risks out of the
fourteen presented (rated at 92.9%, 87.4% and 86.2% respectively). The risk ratings of
writing or reading a text message were almost identical at 79.0% and 78.9%
respectively. This ranked them as the sixth and seventh highest risks. This was only
0.1% lower than the fifth highest risk, speaking on a handheld mobile phone.
Send message
Phone in cradle
Read message
Phone in cradle
Send message
Phone handheld
Read message
Phone handheld
Legal
Illegal
Not sure
Missing
1
1
1
1
1
2
13
4
10
2
1
14
1
3
13
Published Project Report
TRL
36
PPR 367
Table 3.1 A ranked list of the mean risk ratings assigned to driving behaviours
Rank
Driving Behaviour
Percentage
rating of risk
1 When a driver has been drinking (regardless of amount) 92.9
2 When a driver is racing others 87.4
3 When a driver is tired 86.2
4 Other drivers on the road are acting unsafely 81.6
5 When a driver is talking on their mobile phone (handheld) 79.1
6 When a driver is writing a text message 79.0
7 When a driver is reading a text message 78.9
8 When a driver is angry enough to have road rage 77.2
9 When a driver is speeding 71.8
10 When a driver is in a hurry 71.4
11 When a driver is selecting music while driving 67.7
12 When a driver is inexperienced 65.4
13 When a driver is talking on their mobile phone (handsfree) 61.6
14 When passengers are in the car 49.2
3.10 Subjective effects of Texting on performance
3.10.1 Recall of text messages received when driving
During the texting trial, participants were sent two messages and were prompted in the
questionnaire to recall key facts contained within the messages after completing the
drive. Due to technical difficulties, three participants did not receive either text message
and one participant only received one of the messages. As can be seen in Table 3.2, out
of the text messages which were received (14 for the first text and 13 for the second),
virtually all were recalled correctly (only one participant failed to recall the second
question correctly). The dual-task of reading a text message and driving did not seem to
affect recall of the text messages.
Table 3.2 Post-trial message recall performance
Message 1 Message 2
Frequency Percent Frequency Percent
Correct 14 100% 12 92.3%
Incorrect 0 0% 1 7.70%
Missing 3 4
Total 17 17
Published Project Report
TRL
37
PPR 367
3.10.2 Driving performance in the Texting and Control drives
After completion of both simulator drives, participants were asked to rate their
performance in a number of key measures. Questions asked were:
1. Compared to how you normally drive, how well do you think you drove in the first
motorway section?
2. How easy or difficult was it to drive at 60mph and stay in the centre of the lane
during the curve following section?
3. Compared to how you normally drive on curved roads, how well do you think you
drove during the curve following tasks?
4. Compared to how you normally drive when following other vehicles, how well do
you think you drove in the car following section?
5. How easy or difficult was it to maintain a constant distance during the car
following?
6. How easy or difficult was it to respond to any tones you might have heard?
7. How easy or difficult was it to respond to the red bar stimulus which you might
have observed?
8. Compared to how you normally drive, how well do you think you drove overall?
Participants selected a percentage score, with high scores indicating superior
performance. Range, mean and standard deviations were calculated for all questions and
are displayed in Table 3.3.
Published Project Report
TRL
38
PPR 367
Table 3.3 Descriptive statistics for participants’ perceptions of their
performance in the Texting and Control drives (* Ratings represent
percentages where high values correspond to superior driving performance).
Question Texting Control
Minimum 2 0
Maximum 81 86
Mean 55.76 55.00
1. Compared to how you normally drive, how well
do you think you drove in the first motorway
section?
SD 22.13 23.25
Minimum 0 16
Maximum 67 73
Mean 31.12 51.24
2. How easy or difficult was it to drive at 60mph
and stay in the centre of the lane during the
curve following section?
SD 19.86 20.17
Minimum 0 11
Maximum 60 82
Mean 24.47 42.88
3. Compared to how you normally drive on
curved roads, how well do you think you drove
during the curve following tasks?
SD 18.14 23.66
Minimum 0 10
Maximum 70 86
Mean 45.00 46.12
4. Compared to how you normally drive when
following other vehicles, how well do you think
you drove in the car following section?
SD 18.84 23.89
Minimum 0 10
Maximum 65 74
Mean 33.29 34.41
5. How easy or difficult was it to maintain a
constant distance during the car following?
SD 20.26 17.90
Minimum 4 14
Maximum 82 100
Mean 44.82 65.59
6. How easy or difficult was it to respond to any
tones you might have heard?
SD 25.73 23.25
Minimum 0 36
Maximum 79 99
Mean 41.67 66.12
7. How easy or difficult was it to respond to the
red bar stimulus which you might have
observed?
SD 23.57 17.20
Minimum 1 16
Maximum 60 76
Mean 34.94 54.94
8. Compared to how you normally drive, how well
do you think you drove overall?
SD 15.03 18.01
Paired-samples t-tests were performed on the data, comparing participant’s perception
of each measure of performance in the texting whilst driving trial and the control trial,
the results of which are displayed in Table 3.4.
Published Project Report
TRL
39
PPR 367
Table 3.4 Differences in participants’ perceptions of their performance in the
Texting and Control drives (paired samples t-tests)
Question Mean SD t df
Sig.
(2-
tailed)
1. Compared to how you normally drive, how well
do you think you drove in the first motorway
section?
0.76 32.17 0.10 16 0.92
2. How easy or difficult was it to drive at 40mph
and stay in the centre of the lane during the
curve following section?
-20.12 27.77 -2.99 16 0.01
3. Compared to how you normally drive on curved
roads, how well do you think you drove during
the curve following tasks?
-18.41 27.86 -2.72 16 0.01
4. Compared to how you normally drive when
following other vehicles, how well do you think
you drove in the car following section?
-1.12 25.00 -0.18 16 0.86
5. How easy or difficult was it to maintain a
constant distance during the car following? -1.12 15.71 -0.29 16 0.77
6. How easy or difficult was it to respond to any
tones you might have heard? -20.76 26.43 -3.24 16 0.01
7. How easy or difficult was it to respond to the
red bar stimulus which you might have
observed?
-22.67 25.75 -3.41 16 <0.01
8. Compared to how you normally drive, how well
do you think you drove overall? -20.00 20.30 -4.06 16 <0.01
As can be seen in Table 3.4, several significant differences were identified. Participants
rated their performance as significantly worse in the Texting drive for:
• Maintenance of lane position and speed in the curve following section.
• Responding to the RT tasks
• Overall performance
This suggests participants felt their ability to control their vehicles lateral lane position
was affected in the Texting drive (t(16) = -2.99, p < 0.01). However, just as revealing
they did not feel it affected their ability to maintain a constant distance to the vehicle in
front (t(16) = -0.029; p =0 .86). Finally, participants reported that responding to the RT
stimuli was more difficult in the trial (Auditory: t(16) = -3.24, p < 0.01; Visual; t(16) =
-3.41, p < 0.001).
3.10.3 Differences in performance when sending or receiving a text message
Participants were asked how they felt a variety of driving behaviours were affected by
either sending or receiving a text message whilst driving. Specifically, their impressions
of the relative difficulty of sending or receiving a text message across a range of
performance aspects.
Published Project Report
TRL
40
PPR 367
3.10.3.1 Concentration required and keeping in lane
In the first section, two questions asked participants how much concentration was
required when sending or receiving a text message (“virtually no concentration” to
“complete concentration”) and how difficult they found it to stay in a lane when sending
or receiving a text message (“maintained normal positioning” to “struggled to maintain
normal lane positioning”). The results of the questions can be seen at the top of Table
3.5.
Table 3.5 Differences in participants’ perceptions of performance impairment
when sending or receiving a text message.
N Min. Max. Mean SD
Concentration Sending
17 10 86 59.94 22.32
(High scores represent increased
concentration)
Receiving
16 16 80 41.50 17.12
Keeping in lane Sending
17 12 100 70.58 21.94
(High scores represent struggling to
maintain lane position)
Receiving
16 7 94 54.25 19.10
Speed Sending
15 4 81 39.53 21.16
(High scores represent driving faster)
Receiving
16 10 54 38.56 12.77
Distance Sending
17 7 78 36.24 18.57
(High scores represent leaving less
space to the car in front)
Receiving
15 12 94 44.20 18.58
Awareness of hazards Sending
17 41 100 70.41 15.07
(High scores represent being less
aware)
Receiving
16 44 86 62.50 10.99
General driving
performance Sending
17 48 100 76.11 13.23
(High scores represent worse driving
performance)
Receiving
16 45 84 63.37 12.76
From these we can see that participants thought more concentration is required and
keeping in lane is more difficult when sending a message. Paired comparisons t-tests
were performed and the results confirmed that both these differences were significant
(t(15) = 2.50; p = 0.02 for concentration and t(15) = 4.68, p < 0.001 for Keeping in
lane). Figure 3.22 was displays these results.
Published Project Report
TRL
41
PPR 367
0
20
40
60
80
100
Sending Receiving
Concentration
Perceived concentration
required (%)
Higher
Lower
0
20
40
60
80
100
Sending Receiving
Keeping in lane
Perceived difficulty (%)
Higher
Lower
Figure 3.22 Participants’ mean ratings of performance effects of sending or
receiving a text message on Concentration and Keeping in lane
3.10.3.2 Speed selection, distance to vehicle in front, awareness of road hazards and
general driving performance
The four questions in the second section of the questionnaire asked participants to rate
their performance when sending or receiving a text message. However, the scales
allowed participants to report if aspects of their driving were affected positively,
negatively, or were unaffected by sending or receiving a text message. Therefore, the bi-
directional nature of these questions provides us with the answers to two questions:
firstly, does sending impact performance more than receiving (or vice versa) and,
compared to normal driving, how does performance change when sending or receiving a
text message?
The mean, range and SD of each question are listed in Table 3.5. From this table we can
see that there were differences in the participant’s perceived level of performance across
the various behaviours, both between sending and receiving and in comparison with
normal driving behaviour (which is represented by a value of 50). Figure 3.23 illustrates
these differences.
Published Project Report
TRL
42
PPR 367
0
10
20
30
40
50
60
70
80
90
100
Sending Receiving
Speed
Perceived speed change (%)
Higher
Lower
No change
0
10
20
30
40
50
60
70
80
90
100
Sending Receiving
Distance
Perceived distance to lead vehicle (%)
Less
More
No change
0
10
20
30
40
50
60
70
80
90
100
Sending Receiving
Hazards
Perceived awareness of hazards (%)
Less
More
No change
0
10
20
30
40
50
60
70
80
90
100
Sending Receiving
General performance
Perceived general performance (%)
Worse
Better
No change
Figure 3.23 Participants’ mean ratings of performance effects of sending or
receiving a text message on Speed, Distance, Hazard awareness and General
performance
Initial inspection revealed that there was virtually no difference in participants’
subjective appraisal of speed choice when sending or receiving a text message.
However, both figures appeared to be substantially lower than the ‘no change’ level of
50%. Independent samples t-tests were computed for sending and for receiving, to
compare to the ‘no change’ level. The comparison for Receiving was significant (t(15) =
-3.58, p < 0.001) whilst the result for Sending neared significance (t(14) = -1.92; p =
0.08). This demonstrates that participants reported selecting driving more slowly than
normal when receiving a text messaging.
Figure 3.23 seems to display a larger difference in the ratings of distance to the vehicle
in front. The figure shows that participants reported leaving more distance when Sending
than Receiving a message. However, the comparison only neared significance (t(14) = -
1.78; p = 0.10). Although both results are below the ‘no change’ value, independent
samples t-tests demonstrated that a significant result was only achieved for Sending a
message (t(16) = -3.06; p < 0.01). This suggests that participants did leave more
distance to the vehicle in front than they usually would when sending a text message.
Drivers reported being less aware of hazards when Sending than when Reading a
message and a paired-samples t-text confirmed this as a significant difference (t(15) =
3.28; p < 0.01). To determine if these differences were significantly different from the
‘no change’ level, independent samples t-tests were computed for both sending and
receiving, both of which were significant at the 0.001 level (Sending: t(16) = 5.58; p <
0.001, Receiving: t(15) = 4.55; p < 0.001). Therefore, participants were less aware of
hazards when both sending and receiving a text message, and their awareness was
worse when sending.
Finally, participants also rated their general performance. As with the hazard awareness
results, both sending and receiving lead to worse general performance than normal
(Sending: t(16) = 5.58; p < 0.001; Receiving: t(15) = 4.55; p < 0.001). Additionally,
Published Project Report
TRL
43
PPR 367
Sending produced a greater reduction in general performance than Receiving, confirmed
by a paired-samples t-test (t(15) = 3.08, p < 0.01).
3.11 Personality tests
Participants completed extracts of the IPIP relating to three personality measures: Self-
efficacy; Openness to values; and Impulsiveness. They also completed the DBQ (Reason
et al., 1990). Unfortunately, the correlations between overall scores on the
questionnaires and both the driving behaviour measures and the other questionnaire
measures failed to reveal any significant results. This is perhaps unsurprising since it
was relatively ambitious to try to find significant results using personality tools with such
a small sample.
Published Project Report
TRL
44
PPR 367
4 Discussion
This study sought to identify the impairment to driving behaviour caused by concurrent
participation in a range of text message tasks. It followed from research findings by the
RAC Foundation (RAC Foundation, 2008) that significant numbers of young drivers
reported writing and reading text messages whilst driving. Young male and female
drivers were recruited to participate in the study. They were required to drive a high
fidelity driving simulator through the same test scenario twice. This enabled a potentially
dangerous task to be completed repeatably and in complete safety. In one instance,
participants completed a number of text message tasks and in the other, they drove
undistracted.
The results demonstrated that participants’ driving behaviour was impaired by
concurrent text message tasks. Writing text messages created a significantly greater
impairment than reading text messages. Behaviour in response to the arrival of an
ignored text message was unaffected.
Reaction times to (task-unrelated) trigger stimuli tended to be higher when reading or
writing a message. This corresponds with the results reported in Burns et al. (2002) who
showed that drivers’ reaction times were significantly higher with concurrent mobile
phone conversations (using either handheld or handsfree phone). The slowest average
reaction time was observed for drivers responding to the visual reaction time task whilst
trying to compose a text message where reaction times increased from 1.2 to 1.6
seconds. Furthermore, participants were significantly more likely to fail to respond to the
reaction time stimuli if engaged in concurrent text messaging. The failure to detect
hazards and increased response times to hazards has clear implications for safety. At
motorway speeds (as were present in the visual RT task), the increase in mean reaction
time would result in an increased stopping distance of 12.5m (approximately three car
lengths). This could easily make the difference between causing and avoiding an
accident or between a fatal and non-fatal collision. The average completion time for
these rather simple messages was over one minute (nearly three times longer than
when undistracted). On a motorway, a car driver may have travelled more than one mile
with impairment.
It was observed that drivers tended to reduce their speed in the texting conditions. This
corresponds with the results of Kircher et al. (2004) who found that participants tended
to reduce their speed when receiving text messages. Wilde (1982, 1988, 1994)
described this phenomenon as ‘risk homeostasis’ whereby in response to a change in the
road-vehicle-user system, behaviour changes to maintain a target level of risk per unit
time. Whilst there is some debate about the exact processes involved (see Grayson,
1996), the evidence from the questionnaire supports the theory that drivers were aware
that their driving was impaired to some degree whilst engaged in text messaging tasks
and chose to reduce their speed in order to mitigate accident risk.
The most conspicuous change in performance was that observed when texting in the
loops section. The overall pattern revealed large increases in variability of lane position
resulting in many more lane departures when texting. The Department for Transport
STATS19 figures for 2006 reveal that 15% of all accidents and 35% of fatal accidents
were due to loss of control, highlighting the risk that drivers face when their driving is
impaired due to concurrent texting. It was further identified that the impairment caused
by texting was far more significant for female than male drivers. It is possible that this
reflects classic sex differences in motor skills requiring accurate targeting and finger
dexterity (see Kimura, 1999). The survey by the RAC Foundation (2008) found that male
drivers constituted the majority of the “Multi-tasking multimedia maestros” who were
more likely to text and drive. Consequently, although male drivers may show a reduced
impairment when texting and driving, the increased probability that male drivers will
engage in this behaviour suggests that the overall impairment across the sexes may be
more evenly spread. The results found in this study are based on a small sample of
drivers and it would interesting to study this phenomenon further.
Published Project Report
TRL
45
PPR 367
It was observed that participants, when texting, were less able to maintain a constant
distance behind a lead vehicle and showed increased variability in lateral lane position
when following that vehicle. This has significant implications for a driver’s ability to
control a vehicle in normal traffic flows. In the simulated situation, the lead vehicle
varied its speed in a relatively benign manner so collision risk was insubstantial. The lack
of any other traffic in that part of the task meant that participants could afford to leave
large safety margins whilst texting. However, in a normal, real world traffic situation,
vehicles following the driven vehicle would create pressure for a texting driver to
maintain progress relative to vehicles ahead, which in turn may decelerate more rapidly
than was experienced in the simulator scenario. Poorer control of vehicle speed, lateral
position, and increased reaction times in this situation dramatically increase the
likelihood of collision.
As hypothesised, reading text messages had a less detrimental effect on performance
but a detrimental effect nevertheless. Reaction times were slower and lane position was
more variable than under control conditions. There was also an indication that drivers
reduced speed when reading messages, suggesting that they recognised the impairment
to driving ability caused by trying to read the text message and drive. This pattern of
results is consistent with a lower relative task demand of reading a text message
compared to writing a message where, in addition to viewing the phone display screen,
the driver has increased cognitive load when considering message composition and
increased physical load due to greater interaction with the phone keypad.
No changes in behaviour were observed when drivers were required to ignore a text
message that they received whilst driving. This suggests that, if drivers can resist the
temptation to read a received message, there is little harm in a driver leaving their
phone switched on. Indeed, text messaging can be a very useful technique for
communication with a driver, provided the driver chooses to stop in a safe place when
they decide to read/write a message.
The questionnaire results indicated that participants were familiar with the operation of
their phones and tended to leave them active when driving, typically not in a suitable
cradle. There was some confusion about the legality of texting whilst driving. The
majority of participants correctly reported that use of a phone for texting whilst handheld
is illegal. However, a majority of participants also incorrectly reported that use of a
phone for texting whilst in a cradle is illegal. Furthermore, a majority of participants felt
that use of a phone for texting whilst in a cradle should be illegal. This study did not
investigate the effect of texting on driving performance when the phone was in a cradle.
However, the observation that participants brought the phone as near as possible to
their eye-line when driving to read/write messages suggests that separation of the driver
from the phone display (by locating it in a cradle) would impair driving performance by
at least as much a that observed here.
The subjective assessments of performance suggest that participants had insight into the
observed impairment caused by the various text messaging tasks in the simulator drive.
Participants recognised that, when engaged in a text messaging task, they had poorer
lane positioning, chose to drive more slowly, and kept larger safety margins. They also
recognised that writing/sending a message was more of a distraction than reading an
incoming message. Participants tended to rate texting tasks as subjectively more
dangerous than many other in-car activities. The personality tests that were
implemented did not reveal any significant results; probably due to the lack of statistical
power.
4.1 Comparison with previous distraction studies
Earlier studies at TRL have used a similar methodology to that applied in the current
study to:
Published Project Report
TRL
46
PPR 367
• benchmark the relative performance impairment of mobile phone conversations
against that caused by alcohol consumption to the legal limit (Burns et al. 2002)
• Investigate the influence of cannabis on driving (Sexton, Tunbridge, Brook-
Carter, Jackson, Wright, Stark, & Englehart, 2000)
• Investigate the influence of cannabis and alcohol on driving (Sexton, Tunbridge,
Board, Jackson, Wright, Stark, & Englehart, 2002).
This allows comparisons of the relative impairment caused by texting whilst driving to be
made.
4.1.1 Reaction times
All three previous studies used reaction time tasks to assess relative impairment. The
trigger stimuli and response mechanisms were slightly different to those used in the
current study but bear comparison. Burns et al. found that reaction times were
significantly higher in each of the three test conditions than in the control condition
(12.4% higher when at the legal alcohol limit; 26.5% higher whilst talking on a
handsfree phone; 45.9% higher whilst talking on a handheld phone). Sexton et al.
(2000) found reaction times were 21% higher when drivers were under the influence of
cannabis. In the current study, there was a mean increase in reaction time of 34.7% to
the visual stimulus making it apparently worse than alcohol, cannabis, and handsfree
conversations but less detrimental than using a mobile phone for handheld
conversations.
4.1.2 Speed
In the current study, it was observed that participants drove more slowly in the Texting
drive. Direct comparisons with Burns et al. 2002, Sexton et al, 2000, and Sexton et al.
2002 are not simple because their studies were investigating long lasting performance
impairments. The texting episodes in this study were short (typically around 60 seconds)
so the duration over which speed could fall to a lower level was reduced. However in
Write 1, mean speeds were 5.7% lower and in Write 4, mean speeds were 6.9% lower
(note these were in sections of the drive where speed was unconstrained by traffic or
route configuration). In Burns et al., it was found that participants drove around 2.2%
slower when using their phone handsfree and 4.8% slower when using their phone
handheld (participants drove slightly faster when at the legal limit of alcohol). This
suggests participants feel that they have to compensate for a greater perceived
behavioural impairment caused by texting whilst driving than that caused by talking
whilst driving.
In the studies of cannabis, greater speed reductions were observed when drivers were
under the influence of the drug (7.7% reduction in speed in Sexton et al. 2000; 9.1%
reduction in speed in Sexton et al. 2002). This suggests that the combined physiological
and psychological effect of cannabis caused a greater perceived behavioural impairment
than texting whilst driving.
4.1.3 SDLP
Variation in lateral lane position in is a common measure of driving performance and was
used in each of the three previous studies to assess relative impairment. In each, a
similar set of loops was used and lane keeping ability was assessed in the impaired and
control conditions. In comparing, results from the current study, it is worth noting that
measures of SDLP in each texting episode are likely to be higher than in the previous
studies since the texting episodes are shorter. This may have had a significant effect on
the percentage differences observed.
Published Project Report
TRL
47
PPR 367
Burns et al. found that there were no significant differences in SDLP in the handheld or
handsfree conversation conditions. Driving whilst at the legal limit of alcohol
consumption did result in significantly less steady lane keeping than any of the other
conditions in the study.
Sexton et al. (2000) found that drivers displayed an increase of around 35% in SDLP
with high does of cannabis whilst Sexton et al. (2002) found an approximate 14%
increase in SDLP for the cannabis and cannabis + alcohol conditions.
In the current study, the Read 2 and Write 2 tasks were in presented in the loops
section. Each showed an increase in SDLP, significantly so in the Write 2 task (Read 2:
SDLP increased by 12.7%; Write 2: SDLP increased by 91.4%). Increases in SDLP for
Read 2 and Write 2 were also accompanied a significantly greater number of lane
departures in each task.
It would appear that the combination of increased mental workload required to write a
text message, the control impairment caused by the physical act of holding the phone,
and the visual impairment caused by continually shifting visual orientation between the
phone display and the road ahead resulted in significantly impaired ability to maintain
safe road position, particularly when driving through the loops section. Participants’
reduction in speed indicated their awareness of the impairment caused by texting whilst
driving. However, this attempt to mitigate risk cannot fully compensate for the
deterioration in performance when attempting to text and drive.
Published Project Report
TRL
48
PPR 367
Acknowledgements
The work described in this report was carried out in the Human Factors and Simulation
group of the Transport Research Laboratory. The authors are grateful to Andrew Parkes
who carried out the technical review and auditing of this report.
References
Allen, R.W. & Stein, A.C. (1987) The driving task, driver performance models and
measurements, Proceedings, Second International Symposium on Medicinal Drugs and
Driving Performance, Maastricht, The Netherlands, 1987.
Brookhuis K.A., de Vries, G., de Waard D. (1991) The effects of mobile telephoning
on driving performance. Accid Anal Prev 1991;23:309-316.
Brown, I.D. & Poulton, E.C., (1961). Measuring the spare mental capacity of car
drivers by a subsidiary task. Ergonomics, 4:1, 35040.
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 (TRL
Report RL547). Berkshire, United Kingdom: TRL Limited.
Chisholm, S.L., Caird, J.K., & Lockhart, J. (2007). The effects of practice with MP3
players on driving performance. Accident Analysis and Prevention, 40, 704-713.
Duncan, B. (1995) Calibration trials of TRL driving simulator. TRL Staff Paper
(PA/3079/95) Crowthorne: Transport Research Laboratory (TRL), 1995. Presented at the
Vision in Vehicles VI Conference, 13-16 September 1995, University of Derby, UK
Ginsburg, K.R., Winston, F.K., Senserrick, T.M., Garcia-España, F., Kinsman, S.,
Quistberg, D.A., Ross, J.G., Elliot, M.R. (2008). National young-driver survey: Teen
perspective and experience with factors that affect driving safety. Pediatrics, 121, 5,
1391-1399.
Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger,
C. R., & Gough, H. C. (2006). The International Personality Item Pool and the future of
public-domain personality measures. Journal of Research in Personality, 40, 84-96.
Gras, M.E., Cunill, M., Sullman, M.J.M., Planes, M., Aymerich, M. & Font-Mayolas,
S. (2007). Mobile phone use while driving in a sample of Spanish university workers.
Accident Analysis and Prevention, 39, 347-355.
Grayson, G. B. (1996). Behavioural adaptation: A review of the literature. TRL Report
254. Transport Research Laboratory, Crowthorne.
The GSM Association (2007). 20 Year Fact Sheet. Available from:
http://www.gsmworld.com/documents/20_year_factsheet.pdf [Accessed 9 June 2008]
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. Accident Analysis and Prevention, 38, 185-191.
Hosking, S., Young, K., & Regan, M. (2006). The effects of text messaging on young
novice driver performance. Monash University Accident Research Centre, Report No.
246.
Just, M.A., Keller, T.A. & Cynkar, J., (2008). A decrease in brain activation
associated with driving when listening to someone speak. Brain Research, Apr 18, 1205,
70-80
Kimura, D. (1999). Sex and cognition. Cambridge, MA: MIT Press
Published Project Report
TRL
49
PPR 367
Kircher, A., Vogel, K., Bolling, A., Nillson, L., Patten, C., Malmstrom, T., & Ceci,
R. (2004). Mobile telephone simulator study. Swedish National Road and Transport
Research Institute, Linkoping, Sweden.
Lesch, M.F. & Hancock, P.A. (2004). Driving performance during concurrent cell-
phone use: aare drivers aware of their performance decrements? Accident Analysis and
Prevention, 36, 471-480.
McEvoy, S.P., Stevenson, M.R., & Woodward, M. (2006). Phone use and crashes
while driving: a representative survey of drivers in two Australian states. The Medical
Journal of Australia, 185 (11/12): 637-641
Mobile Today (2008) Nokia strengthens its grip with 43% of UK market.
http://www.mobiletoday.co.uk/Nokia_43_percent_share_UK_market.html [Accessed
03/09/2008].
Parkes, A.M., Luke, T., Burns, P.C. & Lansdown, T. (2007). Conversations in cars:
the relative hazards of mobile phones. TRL Report TRL664, Crowthorne, England:
Transport Research Laboratory.
The RAC Foundation (2008). Almost half of Britain’s motorists TXT + DRV.
http://www.racfoundation.org/index.php?option=com_content&task=view&id=530&Item
id=35 [Accessed 10 June 2008]
Reason, J. T., Manstead, A. S. R., Stradling, S., Baxter, J., Campbell, K. (1990).
Errors and violations on the roads. Ergonomics, 33, 1315-1332.
Redelmeier, D.A., Tibshirani, R.J. (1997) Association between cellular-telephone
calls and motor vehicle collisions. N Engl J Med 1997;336:453-8.
Sexton, B.F. (1997) Validation trial for testing impairment of driving due to alcohol.
TRL Report 226, Crowthorne, England: Transport Research Laboratory.
Sexton, B.F., Tunbridge, R.J., Brook-Carter, N., Jackson, P.G., Wright, K., Stark,
M.M., Englehart, K. (2000) The Influence of Cannabis on Driving. TRL Report 477,
Crowthorne, England: Transport Research Laboratory.
Sexton, B.F., Tunbridge, R.J., Board, A., Jackson, P.G., Wright, K., Stark, M.M.,
and Englehart, K. (2002). The Influence of Cannabis and Alcohol on Driving. TRL
Report 543. Crowthorne, England: Transport Research Laboratory.
Stationery Office, The (2007) The Official Highway Code, The Stationery Office Books,
Norwich, England.
Strayer, D.L., Drews, F.A. & Crouch, D.J. (2006). A comparison of the cell phone
driver and the drunk driver. Human Factors, 48, 2, pp 381-391.
Telecoms Market Research (2008) UK Mobile operator market share data.
http://www.telecomsmarketresearch.com/resources/UK_Mobile_Operator_Subscriber_St
atistics.shtml [Accessed 03/09/2008].
Thulin, H., & Gustafsson, S. (2004). Mobile phone use while driving: Conclusions
from our investigations. VTI rapport 490A. Swedish National Road and Transport
Research Institute.
Wilde, G. J. S. (1982). The theory of risk homeostasis: implications for safety and
health. Risk Analysis vol 2, pp 209-225
Wilde, G. J. S (1988). Risk Homeostasis Theory and Traffic Accidents-Propositions,
Deductions and Discussion of Dissension in Recent Reactions. Ergonomics, 31 (4), 441-
468.
Wilde, G. J. S. (1994). Target Risk. PDE Publications, Toronto.
Published Project Report
TRL
50
PPR 367
Appendix A The TRL Driving Simulator Centre
A.1 TRL Driving Simulator
TRL has successfully operated a driving simulator for more than 15 years and in that
time the simulator has seen a number of different incarnations to keep pace with
improvements in vehicle, projection, computing, and simulation technologies and as such
is one of the most advanced simulators in the UK. The latest iteration uses a Honda Civic
family hatchback (see Figure A.1). Its engine and major mechanical systems have been
replaced by a sophisticated electric motion system that drives rams attached to the axles
underneath each wheel. These impart limited motion in three axes (heave, pitch, and
roll) and provide the driver with an impression of the acceleration forces and vibrations
that would be experienced when driving a real vehicle. This significantly enhances the
realism with which drivers approach the driving task and reduces the incidence of
simulator sickness (a condition with symptoms similar to those of motion sickness)
among participants. All control interfaces have a realistic feel and the manual gearbox
can be used in the normal manner (automatic gears can be simulated).
Figure A.1 TRL driving simulator, CarSim
Surrounding the simulator vehicle are large display screens onto which are projected the
graphic images that represent the external visual environment to the driver. The level of
environmental detail includes photo-realistic images of buildings, vehicles, signing, and
markings, with terrain accurate to the camber and texture of the road surface. We have
also recently added the capability to simulate night-time driving scenarios. The driving
environment is projected at a resolution of 1280×1024 onto three forward screens to
give the driver a 210º horizontal forward field of view. The presence of the two flat side
screens adjacent to the driver gives a very strong impression of other vehicles travelling
alongside of the vehicle. A rear screen provides a 60º rearward field of view, thus
enabling normal use of all mirrors.
Published Project Report
TRL
51
PPR 367
Surveillance video cameras are mounted in the car and participants can be recorded
during their drive. There is also an intercom facility for communication between the
vehicle and the control room. An in-car colour LCD display can also be used to give
instructions or provide other task-related information.
Figure A.2 TRL CarSim: Control Room
More than one hundred autonomous traffic vehicles can be programmed to participate in
the simulation. TRL has a library of different vehicle types to choose from including cars,
trucks, buses, emergency vehicles, bicycles, and pedestrians. Each obeys specific driving
rules to behave in a normal manner with respect to other traffic vehicles. However,
these can be overridden causing them to perform specific manoeuvres e.g. emergency
stop, sudden lane change etc. The autonomous vehicles also have dynamic properties of
their own – they appear to pitch realistically under acceleration and braking, and vehicle
graphics include body tilt and roll under braking, acceleration and turning; speed
dependent rotating wheels and fully working brake, indicator, fog, and head lights. These
provide additional cues to the driver and greatly enhance the realism of a scene. To
generate scenarios with a heavy traffic load (> 1700 vehicles per lane per hour) we can
generate a vehicle 'swarm'. The swarm function allows us to define a region around the
driver where vehicles will be placed and controlled. A vehicle moving out of the visible
range of the driver is replaced by a new vehicle positioned to maintain the desired traffic
density. This gives the impression of very high volume of traffic while maintaining the
performance of the simulator.
A stereo sound system with speakers inside and outside the vehicle generates realistic
engine, road, and traffic sounds to complete the representation of the driving
environment. The software used to implement the simulation is called SCANeR II and
was created by OKTAL to provide a flexible and powerful simulation with a highly
advanced traffic model. It is employed by more than twenty research institutes across
the globe and TRL leads the user group with access to OKTAL expertise for trial set-up
and integration, if required.
The dynamics of the vehicle are modelled using a validated vehicle model that is used for
product development by Renault. The model interprets the driver’s control inputs, relates
Published Project Report
TRL
52
PPR 367
them to the current vehicle status and computes a prediction of how a real vehicle would
behave in the given circumstances. The system then responds to present to the driver its
optimal representation of how this behaviour would be perceived through the visual,
sound, and motion sub-systems. The vehicle dynamics are updated at 100Hz whilst the
visuals are refreshed at 60Hz so that the driver perceives a seemingly continuous driving
experience. Data is then recorded relating to all control inputs made by the driver,
including steering, pedals, gear, indicators; vehicle parameters such as speed, RPM; and
parameters to assess behaviour in relation to other vehicles such as distance and time
headways. The data recording rate is fully controllable dependent upon the trial
demands, up to a rate of 100Hz.
The simulator also includes a full integrated SmartEye eye–tracking system for the
analysis of driver visual behaviour. This system, in addition to being able to report the
driver’s gaze direction, is integrated with the 3D environment presented in the
simulation, such that the eye-tracker can report in the simulator data the specific
element on which the participant is fixating – a specific road sign, traffic light, the road
ahead, or interior items such as the instrument panel or infotainment system. This
dramatically improves the accuracy and efficiency of post-trial data analysis.
Participants for trials are recruited from a dedicated database of over 1000 members of
the public. This comprises drivers from a wide range of ages and backgrounds, all of
whom are familiar to TRL such that participants from particular demographic bands or
driving experience/ability ratings can be selected to suit the trial requirements. The
simulator facilities include a medical room for taking any physiological measures and
trials management staff are trained in Good Clinical Practice. There is an interview room
for questionnaire completion and debriefing and an information room for conducting
computer based test or training tasks. Data management procedures are well
established and compliant with the Data Protection Act 1998 to ensure security,
confidentiality, and integrity of all records.
Published Project Report
TRL
53
PPR 367
Appendix B Phone makes and network providers
Error! Reference source not found. shows the count of makes of participants’ mobile
phone used in the study and network providers used by each participant.
Samsung; 4
Sony Ericsson; 4
Nokia; 8
Sagem; 1
Orange; 4
3G; 1
Vodafone; 3
Tmobile; 1
O2; 8
(a) (b)
Figure B.4.1 Count of (a) mobile phone makes and (b) network providers used
in the study
Given the relatively small sample size the distribution of mobile phone makes does
match the UK market share of each manufacturer reasonably well (August 2008: Nokia
43%; Sony Ericsson 25%; Samsung 21% (Mobile Today, 2008)). One notable absence
from the study was the manufacturer, Motorola, although that company is experiencing
a fall in market share currently. The distribution of network providers among participants
is less consistent with UK market share. In the UK, phone users are spread
approximately equal across the ‘big four’ (Q1 2008: Vodafone 26.6%; O2 26.4%; T-
mobile 24.5%; Orange 22.6% (Telecoms Market Research, 2008)). The skewed
distribution amongst participants may be due to differences in signal strength in the local
area or may simply be a function of the small sample size.
Eight participants reported using predictive text for message composition (four male;
four female). Nine participants did not use predictive text (five female; four male).
Published Project Report
TRL
54
PPR 367
Appendix C Messages sent in Texting drive
Table C.1 shows the text messages composed by each participant in the Texting drive.
Table C.1 Text messages composed by each participant in their Texting drive
Participant number
Task
1 2 3 4 5 6 7 8 9 10
11
12
13
14
15
16
17
Write practice
I am drin a gr8 car
simulator.
I am driving a great car
simulator.
I am driving a great car
simulator
i am driving a great car
simulator
Im driving a great car
simulator
I am driving a great car
simulator
I am driving a great car
simulator!
I am driving a great car
simulator
I am driving a great car
sim.
I am driőing a great car
simulator.
I am driving a great car
simulator
I am driving a great car
simulator
I'm driving a great car
simulator
I an driving a great car
simulator
I am driving a great car
simulator
I am driving a great car
simulator
Write 1
Happy birthday hav fun at
the party
Happy bday.have fun at the
party
Happy bday have fun at the
party
happy birthday have fun at
the party!
Happy birthday have fun at
the party
Happy bday have fun at the
party
Happy birthday,have fun at
the party!
Happy bday. Hav fun at the
party
Happy birthday. Have fun
at the party.
Happy birthday have fun at
the party.
Happy birthday, have fun
at the party.
Happy birthday hav fun at
the party
Happy birthday, have fun
at the party
Happy birthday have fun at
the party
Happy birthday have fun at
the party
Happy birthday, have fun
at the party
Happy birthday have fun at
the party
Write 2
Nice 2cu at the cafe
yesterday
Nice to see u at the cafe
yesterday
nice to see u at the cafe
yesterday
Nice to see u at the cafe
yesterday
Nice 2 c u at the cafe
yesterday
Nice c see u at the cafe
yesterday!
Nice to c u at the cafe
yesterday
Nice to see u at the cafe
yesterday.
Niåe to se u at the cafe
y'day
Nice to see you at the cafe
yesterday
Nice 2 c u at the cafe
yesterday
Nice to see you at the cafe
yesterday
Nice to c u at the cafe
yesterday
Nice to see you at the cafe
yesterday
Nice to see you at the cafe
yesterday
Nice to cu at the cafe
yesterday
Write 3
Dont worry have a nice
time in paris
Dont worry, have a nice
time in paris
Dont worry have a nice
time in paris
dont worry have a nice
time in paris
Dont worry, have a nice
time in paris
Dont worry, have a nice
time in paris
Dnt worry have a nice time
in paris
Dnt worry hav a nice time
in paris
Don't worry. Have a nice
time in paris.
Doot worry have a gd time
in paris
Don't worry, have a nice
time in paris!
Dont worry have a nice
time in paris
Don't worry have a nice
time in paris
Don't worry have a nice
time in paris
Don't worry have a nice
time in paris
Dont worry have a nice
time in paris
Write 4
Sorry bout ur ankle get well
soon
Sorry about your ankle.get
well soon
Sorry about your ankle get
well room
sorry about ure ankle get
well soon
Sorry bout ur ankle, get
well soon
Sorry bout ur ankle get well
soon
Sorry about ur ankle, get
well soon!
Sorry about ur ankle get
well soon
Sorry about your ankle. Get
well soon.
Sorry abovt ur ankle get
well soon.
Sorry about your ankle, get
well soon
Sorry about ur ankle get
well soon
Sorry about your ankle, get
well soon
Sorry about your ankle, get
well soon
Sorry about your ankle. Get
well soon
Sorry about youre ankle
gdu well soon
