You are on page 1of 12

BEHAVIOUR & INFORMATION TECHNOLOGY, 2017

VOL. 36, NO. 4, 435–445


http://dx.doi.org/10.1080/0144929X.2016.1240234

Texting and walking: a controlled field study of crossing behaviours and


inattentional blindness in Taiwan
Ping-Ling Chena, Wafaa Salehb and Chih-Wei Paia
a
Graduate Institute of Injury Prevention and Control, College of Public Health, Taipei Medical University, Taipei City, Taiwan, ROC; bTransport
Research Institute, Edinburgh Napier University, Edinburgh, Scotland

ABSTRACT ARTICLE HISTORY


The paper investigates the effects of phone use (talking, texting, and listening to music) on the Received 10 June 2015
street-crossing behaviours of pedestrians and their inattentional blindness in Taiwan. Recent Accepted 19 September 2016
handsets with touchscreens, as well as more advanced features including multimedia, and
KEYWORDS
mobile applications (apps), exacerbate problems relating to cognitive distraction and reduced Unsafe crossing behaviour;
situation awareness. A controlled field study using video cameras was conducted for observing phone use; texting and
pedestrians’ crossing behaviours (e.g. crossing time, sudden stops, looking both ways before walking; pedestrian safety
crossing, and disobeying traffic signals). Pedestrians were classified into two groups:
experimental group (talking, texting, and listening to music) and control group (no phone use).
Pedestrians’ inattentional blindness was examined by evaluating whether they saw and heard an
unusual object (i.e. a clown) nearby. The results indicate that the proportions of unsafe crossing
behaviours (e.g. sudden stops, disobeying traffic signals, and not looking both ways before
crossing) were higher among distracted individuals and more pronounced among those using
instant-messaging apps. These instant-message app users were the least likely to see the clown,
and music listeners were the least likely to hear the horn that the clown was honking.
Contributing factors to unsafe behaviours include being a student, having a phone screen of
5 inches or larger, and having unlimited 3G Internet access. Texting message via apps was the
leading factor on unsafe crossing behaviours of pedestrians and their inattentional blindness.

1. Introduction
while driving constitutes a dual task that often requires
Considerable research has examined the dangers of using drivers to fixate their eyes for texting on the handsets,
mobile phone while driving. The literature has suggested reducing their ability to control their vehicle, such as
that talking and driving constitutes a dual task that slows through reduced manoeuvrability, increased speed varia-
brake reaction time, reduces lateral and longitudinal bility, and improper lane changes (see a review study by
vehicle control, and causes increased cognitive distraction Caird et al. 2014). As reported by Young et al. (2014)
and reduced situation awareness (Collet, Guillot, and Petit who have examined drivers’ driving performance in a
2010). Simulator studies (Horrey and Simons 2007) have simulated environment, receiving and sending text mess-
concluded that mobile phone users may compensate for ages through smart phones caused decrements in speed
their reduced environmental awareness by driving more monitoring, reduced the amount of time that drivers
slowly and following farther behind other vehicles to pre- looked at the forward roadway, and increased workload.
vent rear-end accidents with the leading vehicles. Although Other studies (Owens, McLaughlin, and Sudweeks 2011)
the simulator studies (Horrey and Simons 2007; Strayer, have reported that texting using smartphones results in
Drews, and Johnston 2003) have offered evidence of more driver errors than does texting using keypad
behavioural compensation, caution should be exercised phones. These dangers are particularly evident for
in generalising such evidence to real-world situations. young drivers who were found to be unlikely to stop
Recent handsets with touch screens and more text messaging tasks in a difficult driving situation
advanced features, including multimedia and mobile (Hosking, Young, and Regan 2009). National statistics
applications (apps), exacerbate the above-mentioned in the USA indicate that texting while driving contrib-
problems. Texting is a particular hazard to all road utes to approximately one million crashes every year
users because it involves multiple types of distraction: (Olson et al. 2009). A Virginia Tech Transportation
manual, visual, and cognitive. For example, texting Institute study investigating the impact of distracted

CONTACT Chih-Wei Pai cpai@tmu.edu.tw


© 2016 Informa UK Limited, trading as Taylor & Francis Group
436 P.-L. CHEN ET AL.

driving in commercial motor vehicle operations reported (Hyman, Sarb, and Wise 2014; Strayer, Drews, and John-
that texting while driving increases the crash risk of a dri- ston 2003). Hyman, Sarb, and Wise (2014) reported that
ver by 23 times (Neider et al. 2010). people talking on a mobile phone while walking failed to
Considering the impact that texting may have on see a unicycling clown. Hyman, Sarb, and Wise (2014)
driving safety, recently concerns have been raised later pointed out that their participants when walking
about how texting may affect pedestrian safety. Results and texting messages were less likely to show awareness
from the studies investigating texting and walking indi- of money in a tree (which serves as an unusual object)
cate that texting creates a significantly greater interfer- beside the pathway on which they were walking. These
ence effect on walking than does listening to music or past studies have attributed inattentional blindness to
talking on a mobile phone. Compared with pedestrians the possibility that divided attention in a complex
who were using their phones for other purposes (e.g. environment reduces awareness of objects that are not
talking or listening to music), pedestrian texters were the focus of attention, and that the objects that people
less able to maintain walking speed, and more likely to fail to see can be interesting and surprising, but are not
pose a threat to public safety by ignoring their surround- directly related to the person’s primary task.
ings (or ambling across the street) (Neider et al. 2010). A Concurrent with other countries that have outlawed
laboratory study by Caird et al. (2013) examined whether mobile phone use (including texting) while driving, Tai-
browsing the Internet on a smartphone while crossing a wan has banned mobile phone use while driving (includ-
virtual street compromised the safety of young adult ped- ing while riding a motorcycle). Official statistics on
estrians. The study reported that distracted pedestrians texting-related pedestrian causalities are few; however,
were found to wait longer to cross the street, missed with the advances in smartphone features and functions,
more safe opportunities to cross, took longer to initiate using phones while driving may remain a frequent prac-
crossing when a safe gap was available, looked left and tice. Followed by Fort Lee in the US state of New Jersey,
right less often, spent more time looking away from where jaywalking pedestrians may be ticketed if caught
the road, and were more likely to be hit or almost hit texting and walking (Apple Daily 2014), Taiwan is devel-
by an oncoming vehicle. oping a law to ban dangerous walking behaviours such as
Research has suggested that texting while walking texting and walking (Kahneman 1973).
affects gait performance (i.e. ability to walk and balance), While reviewed together, the literature has indicated
which may affect the safety of pedestrians. Rubinstein, that texting reduces drivers’ attention to the road
Meyer, and Evans (2001) reported that walking and text- environment and substantially increases the likelihood
ing on a mobile phone may modify gait performance of crashes, injuries, and fatalities. Research that has
because of the increased cognitive demand placed on examined the risks associated with distracted walking
working memory and executive control in performing has also reported that text messaging on smartphones
dual tasks. Lamberg and Muratori (2012) recently modifies gait performance, causes unsafe street-crossing
revealed reduced walking speed and deviation from a behaviours, and leads to inattentional blindness that may
straight path while texting messages. Recently Schabrun increase the likelihood of accidents involving both the
et al. (2014) demonstrated larger decreases in gait vel- pedestrians themselves and other road users. A primary
ocity and greater lateral deviation in young adults walk- reason that distracted pedestrian activity is particularly
ing and texting compared with those walking and talking risky is multitasking; attempting to perform two cogni-
on a phone. tively complicated tasks simultaneously reduces atten-
As Zhou, Horrey, and Yu (2009) suggested, confor- tion to and the performance of one or both tasks
mity is the tendency to follow other individuals’ beha- (Hancock et al. 1999). Distracted behaviour such as
viours, actions, values, and ideas to avoid potential using a mobile device while crossing a street may be
conflict with others. The literature has reported that expected to be dangerous considering the consequences
the greater the cohesion in a group, the greater the of errors while crossing a street and the need for ped-
group pressure to conform in risky behaviours estrians to simultaneously respond to multiple complex
(Zimbardo and Leippe 1991). In a pedestrian walking stimuli. Studies have suggested that pedestrians dis-
behavioural study conducted in China (Hyman et al. tracted by factors ranging from making phone calls to
2010), pedestrians were found to have stronger inten- text messaging may take greater risks than undistracted
tions to cross against a traffic signal when surrounding pedestrians do (Simons 2000). Less researched are the
pedestrians were doing so. effects of calling and text messaging through online ser-
Research has also shown that inattentional blindness vices and apps, which has increased rapidly in popular-
may occur in naturalistic settings caused by mobile ity, particularly among young people, and which
phone conversations while driving and walking uniquely involves multiple types of distraction: manual,
BEHAVIOUR & INFORMATION TECHNOLOGY 437

visual, and cognitive. The rapid development of mobile 2.2. Procedures


phone features and technologies has increased the com-
All participants were assigned into the experimental
plexity of multitasking, thereby undermining pedestrian
group or control group. The participants were assigned
safety. Such mobile phone developments include larger
to ‘experimental’ and ‘control’ groups prior to walking
screen sizes, which may distract users to a greater degree;
the street. There were two types of volunteers in our
third- and fourth-generation (3G and 4G) mobile com-
study: the control type (i.e. non-distracted group) was
munication technology, which offers quicker data trans-
those who were required to come alone. These partici-
mission; and more-advanced instant-messaging apps,
pants served as the subjects in the non-distracted
which offer more convenient means of talking and text
group where the only instruction was to walk through
messaging. To the authors’ knowledge, few studies
the street (i.e. they were not receiving any calls/texts
have identified the effects of smartphone features and
from anyone). The second type of volunteers (i.e. dis-
technologies on the unsafe crossing behaviours or inat-
tracted group) was those who were required to come
tentional blindness of pedestrians. One exception was
with a friend. The friend was asked to make a call or
the laboratory study by Byington and Schwebel (2013);
text message (the calls and texts were made either by a
however, whether laboratory studies accurately reflect
traditional way or an app) to the participant, while the
real-world texting and street-crossing behaviours is
participant was directed to walk through the street.
unknown.
The control group comprised participants who crossed
the street while not distracted at all, and the experimental
1.1. Purpose group comprised the individuals distracted by phone use
or listening to music. Classifying an individual to the
The current research improves upon Byington and experimental group or control group is based on whether
Schwebel (2013) who examined the crossing behaviours an individual was assigned to conduct a distracting
of pedestrians distracted by mobile phone Internet activity prior to walking through the street. That is, an
browsing in a virtual environment, by conducting a individual is assigned to the experimental group if he/
field-controlled study. The main purpose of this study she was assigned to listen to music, make a call, or text
is to investigate the effects of phone use (talking, texting, message. Making a call and texting can be conducted
and listening to music) on pedestrian street-crossing using an app or the traditional method. On the other
behaviours. The present paper also examines the hand, an individual is assigned to the control group if
relationship between distractions (i.e. phone use) and there was no phone use at all. Participants in the exper-
inattentional blindness. iment group can be users of smartphones (for perform-
ing the five distraction events, as shown in Table 1) or
feature phones (for performing listening to music, or
2. Method
talking/texting using the traditional method). Partici-
2.1. Participants pants in the control group can be users of either smart-
phones or feature phones. All participants were naïve to
To be eligible, participants had to be mobile phone users
the purpose of the study; the only instruction they
(either feature phones or smartphone) and be familiar
received was to cross the street. That is, the participants
with text messaging and making phone calls. Participants
were instructed from the lab down to the street and
were ineligible if they reported a history of medical ill-
crossed the street without any reason. Furthermore, par-
ness relating to vision or hearing, or any conditions
ticipants were unaware that they were being video
that restricted their ability to walk, send a text message,
recorded, and were unaware of incoming text messages
or place a phone call. Smartphones can have either
or phone calls that serve as distractions.
touchscreens or keypads. To identify smartphone users,
To examine the effect of listening to music, the par-
only those using smartphones with touchscreens were
ticipants assigned to be music listeners were required
included in the study as keypad smartphones were
to listen to pop music that had been loaded onto their
found to be very rare. In total, 1669 participants who vol-
device beforehand (the same song was used for all par-
unteered to participate in the controlled field study were
ticipants to avoid bias). Determining the effects of differ-
recruited from Taipei Medical University, Taiwan, and
ent music types on unsafe crossing behaviour is out of
the University Hospital. Volunteers were offered a
scope of the current research.
small gift (i.e. a convenience store voucher worth
Regarding the effects of texting and making a call on
approximately US$2) as an incentive. All aspects of the
crossing behaviours, the sender or receiver of the text
study were approved by Taipei Medical University Insti-
message (or the maker of the call), and the content of
tutional Review Board.
438
P.-L. CHEN ET AL.
Table 1. Distribution of various distraction types for independent variables.
Distraction type
Talking (traditional) Talking (using an app) Texting (traditional) Texting (using an app) p-
Characteristics Listening to music (%) (%) (%) (%) (%) None (%) Value
Gender Male 83 (57.2) 125 (56.3) 118 (55.1) 87 (49.4) 94 (54) 438 (60.2) .326
Female 62 (42.8) 97 (43.7) 96 (44.9) 89 (50.6) 80 (46) 290 (39.8)
Occupations Students 93 (76.9) 140 (82.8) 149 (81.9) 84 (75.7) 89 (73.6) 920 (96.3) <.01
Administrators 16 (13.2) 18 (10.7) 23 (12.6) 21 (18.9) 25 (20.7) 19 (2)
Faculty 12 (9.9) 11 (6.5) 10 (5.5) 6 (5.4) 7 (5.7) 16 (1.7)
Ageb m: 27.6; s: 7.4 m: 29.8; s: 13.6 m: 22.7; s: 6.7 m: 25.6; s: 5.7 m: 22.3; s: 5.2 m: 30.7; s: 9.0 <.01
Phone characteristics Smartphone 169 (93.4) 223 (91) 243 (100) 167 (81.9) 195 (100) 348 (47.8) <.01
Feature phone 12 (6.6) 22 (9) 0 (0) 37 (18.1) 0 (0) 380 (52.2)
Screen size <5 inches 83 (57.2) 98 (43.8) 137 (61.4) 110 (63.2) 89 (53.9) 386 (53) .031
≥5 inches 62 (42.8) 126 (56.2) 86 (38.6) 64 (36.8) 76 (46.1) 342 (47)
3G mobile Internet Unlimited use 76 (43.9) 90 (41.9) 126 (68.1) 92 (43.4) 89 (61.8) 354 (48.6) .381
Restricted use 51 (29.5) 62 (28.8) 59 (31.9) 60 (28.3) 55 (38.2) 232 (31.9)
None 46 (26.6) 65 (29.3) 0 (0.0) 60 (28.3) 0 (0) 142 (19.5)
Number of pedestrians crossing Same direction m: 3.6; s: 2.1 m: 2.3; s: 1.7 m: 1.7; s: 1.3 m: 2.2; s: 1.6 m: 3.7; s: 1.2 m: 2.7; s: 1.5 <.01
legallya,b Opposite m: 4.2; s: 3.9 m: 4.3; s: 3.7 m: 3.7; s: 2.0 m: 4.2; s: 3.3 m: 1.5; s: 2.7 m: 1.8; s: 0.9 <.01
direction
Both directions m: 8.3; s: 6.8 m: 7.6; s: 5.5 m: 6.7; s: 4.7 m: 5.9; s: 2.7 m: 4.7; s: 4.0 m: 3.6; s: 2.8 <.01
Number of pedestrians crossing Same direction m: 1.1; s: 0.9 m: 0.6; s: 0.2 m: 0.8; s: 0.4 m: 1.2; s: 0.9 m: 1.3; s: 0.8 m: 1.0; s: 0.7 <.01
illegallya,b Opposite m: 1.4; s: 1.5 m: 1.7; s: 1.5 m: 1.1; s: 1.0 m: 0.9; s: 0.8 m: 1.2; s: 1.1 m: 0.9; s: 0.8 <.01
direction
Both direction m: 2.3; s: 2.0 m: 3.5; s: 2.7 m: 2.9; s: 1.6 m: 1.6; s: 1.1 m: 2.7; s: 2.2 m: 2.5; s: 1.4 <.01
Number of pedestrians waiting on the Same direction m: 7.4; s: 7.5 m: 5.7; s: 4.8 m: 6.8; s: 5.2 m: 9.7; s: 7.3 m: 6.2.; s: 4.1 m: 8.4; s: 6.5 <.01
sidewalka,b Opposite m: 7.9; s: 6.3 m: 9.9; s: 3.8 m: 8.0; s: 7.1 m: 8.9; s: 6.9 m: 9.9; s: 7.3 m: 5.9; s: 3.0 <.01
direction
Both direction m: 13.2; s: 12.7 m: 15.7; s: 13.2 m: 11.2; s: 10.7 m: 15.9; s: 9.6 m: 16.7; s: 11.7 m: 15.3; s: 10.7 <.01
a
At the moment when the participant arrived.
b
The figures represent the mean (m) and the standard deviation (s).
BEHAVIOUR & INFORMATION TECHNOLOGY 439

the message, may be expected to influence phone use, (i.e. before they stepped into the sidewalk) were con-
such as whether to read or reply to the text. A pilot sur- sidered undistracted ones.
vey was administered to determine how the participants The data were collected from May 2012 to December
responded to the texts or calls from research assistants. 2015 on weekdays during three periods of the day: morn-
Participants tended to ignore texts or calls from the ing peak hours (0700–0900), off-peak time (1200–1400),
strangers (i.e. research assistants), and thus texting and afternoon off-peak hours (1600–1800). Observation
back or calling back was rare. As a result, in the current during the evening was avoided because poor visibility
study, those assigned to make calls and texting messages potentially poses a greater risk of accidents to the
were required to come with a friend, and the friend was participants.
instructed to call or text message the participant before Participant crossing behaviours may be expected to
crossing the street. Immediately after the participants change markedly depending on traffic signal phases.
and their friends arrived in the lab for the study, they All participants were instructed down to the street
were isolated and further instructions were provided from the lab, and they faced traffic signals that were
individually. All participants were being monitored by either in red or green randomly. Of the 896 individuals
CCTV, which research assistants were using to instruct in the distracted group, the phases were red and green
the accompanying friend when to make the call or as soon as 215 and 681 individuals had arrived in the
send the text. intersection. Of the 773 participants in the undistracted
Notably, if both the participant and the friend had the group, 163 and 610 faced red or green phases, respect-
same instant-messaging app (e.g. LINE or WhatsApp) ively. The experimenters (research assistants) acted as
installed on their handsets, text messages were sent and safeguards to prevent the participants from stepping
phone calls were made using the specific app. The into oncoming traffic in the event that they decided to
accompanying friends were required to make verbal violate the red phases. These pedestrians’ tasks were
calls via the apps instead of visual calls. Traditional therefore aborted by the experimenters, and their aver-
text message or phone-call methods were used if the par- age crossing times were not measured. All participants
ticipants did not have the same app. The designed text were instructed to cross the signalised two-lane street
message conversation topic was: ‘I forgot our tutor’s/ (see Figure 1) where the automatic pedestrian count-
supervisor’s phone number. Please tell/text me the num- down device was present providing pedestrians with
ber. Thanks!’ Requesting the tutor’s phone number may remaining seconds for crossing the street. The partici-
result in various outcomes, such as participants ignoring pants would firstly have to intersect with the turning
the contact from their friend. Participants who did not vehicles (from left-hand side) and subsequently the
ignore the contact texted the number back to their friend vehicles from the right-hand side that had completed
or called them. Participants who ignored the contact and the U-turn manoeuvre. The speed limit for the street is
proceeded across the street were considered undistracted 20 km/h, and the timing of the pedestrian-light cycles
walkers. is 90 s (65 s for green light and 25 s for red light).
A pilot survey was administered and found that, to be To investigate the possible effects of phone use on inat-
included in the survey, a task (i.e. the friend texting or tentional blindness, participants’ failure to see or hear the
making a call) had to be initiated when the participant clown was examined. A 185-cm-tall student was recruited
was at least 20 m away from the street (assuming an to be the clown, and to avoid being blocked by regular-size
average walking speed of 1.3 m/s and considering trans- cars (but not by larger vehicles such as buses). The clown
mission lag times and the time required for a friend to costume serves as an unusual object to pedestrians. The
type text messages or make a call) and had to be received clown was standing right next to a parking lot where traf-
when the participant was at least 5 m away from the fic may pose a conflict with crossing pedestrians, making
street, thereby ensuring that the participants crossed the clown more likely to appear within the visual field of
the street and engaged in the distracted activities simul- participants (see Figure 1). The clown was pressing the
taneously. Participants who received and completed the horn, which serves as the acoustic stimulus to partici-
activities (i.e. texting or making a call) before executing pants. It is noteworthy that the position of the clown
their crossings (i.e. walking slowly or standing on the was critical in determining whether he appeared within
sidewalk) were still considered distracted participants; the visual field of the participants. However, as in well-
this is because for a cognitive distraction to exist, dis- known studies of inattentional blindness (Hyman, Sarb,
tracted individuals must reacquaint themselves with and Wise 2014; Newcombe 2012; Thompson et al.
their surroundings after completing a task. However, 2013), controlling this variable to such an extent was
participants who completed the activities much earlier beyond the scope of our experimental design.
440 P.-L. CHEN ET AL.

Figure 1. Designated intersection where the participants were observed, with a clown nearby serving as an unusual stimulus.

2.3. Variables considered


three levels: unlimited use (use as much as you want),
The independent variables considered and measured restricted use (pay as you use), and none (no Internet
include the age, gender, and occupation of the partici- is subscribed).
pants; phone types (feature phone or smartphone); In the current research, social conformity variables
screen size; 3G data limit; and social context. These vari- such as the numbers of pedestrians crossing legally,
ables are described in more details as follows. crossing illegally, and waiting were measured for both
Participant attributes include demographic infor- sides of the street. Video clips were used to collect
mation, namely age, gender, and occupations. Age (in these data when a distracted pedestrian was waiting at
years) was the only continuous variable measured. Occu- a red signal.
pation groups were students, administrators, and faculty The outcome measures (i.e. unsafe crossing beha-
members. Only one temporal factor was examined, viours) observed include crossing time, stopping while
namely ‘time of observation’: rush hours (0800–0900 on the pedestrian crossing, looking both ways before
or 1600–1800) or non-rush hours (1200–1400). crossing, disobeying traffic signals, noticing a clown
The present research hypothesises that smartphone (which serves an unusual stimulus), and hearing the
use would be more detrimental than feature phone use horn played by the clown. Head-turning frequency was
to pedestrian walking performance, and that this effect also observed; studies (Hancock et al. 1999) have demon-
would increase with smartphone screen size. The vari- strated that head turning is crucial to navigating streets
able ‘phone type’ thus comprises two levels: smartphone safely. After the experiment, the participants were inter-
(that have touchscreens) and feature phone. The present viewed by the research assistants and asked whether they
study also hypothesises that phone screen size may be saw the clown nearby, and whether they heard the horn
associated with unsafe walking behaviours; the ‘screen from the clown.
size’ is a categorical variable that comprises two levels: The definitions of the outcome variables (i.e. beha-
smaller than 5 inches and 5 inches or above. viours) are as follows. Participants were considered to
Past research, notably Byington and Schwebel (2013), have stopped suddenly if they stopped at some points
has noted that crossing a virtual pedestrian street while while crossing the street. Looking both ways required a
using the Internet on a smartphone compromises ped- participant to exhibit a noticeable turn of the chin left
estrian safety. In the current research, pedestrian walking and right before stepping onto the street. Similarly, to
performances is assumed to vary according to Internet count the head-turning frequency, a turn of the chin
data limit; for instance, eye fixation duration is probably left and right before crossing the street had to be clearly
longer (and thus a lower likelihood of looking both ways identified. Obeying the traffic signals was defined as
before crossing the street) for people with unlimited 3G crossing during a green pedestrian signal (i.e. a walk
Internet access. The variable ‘3G Mobile Internet’ has signal). Those who crossed the street during a red
BEHAVIOUR & INFORMATION TECHNOLOGY 441

phase (i.e. a do–not-cross signal) were considered as hav- analyses. The category ‘feature phone’ was not combined
ing disobeyed the signal. with the category ‘smartphone’ as doing so would
obscure their meaning (only the category ‘smartphone’
was included in the chi-squared analysis). The distrac-
2.4. Analysis and sample size tion types – listening to music, talking (traditional),
All data collected were entered into an Excel spreadsheet, taking (using an app), texting (traditional), texting
and SPSS (version 22; IBM) was used for data analysis. (using an app), and none – varied depending on occu-
Firstly, several unsafe crossing behaviours were firstly pations (p < .01), ages (p < .01), and phone screen size
compared between the experimental and control groups. (p = .031). Table 1 presents the proportions of partici-
The average crossing times and head-turning frequency pant groups for each distraction type. For instance,
(Table 2) were compared using a t-test. The unsafe cross- 57.2% for male participants regarding the distraction ‘lis-
ing behaviours of participants in the experimental group tening to music’ indicates that 57.2% of participants who
were compared with those of participants in the control listened to music during the study were male.
group by using χ 2 post hoc tests. Next, we examined the
determinants of unsafe crossing behaviours in the exper-
iment group. All multivariate analyses were conducted 3.2. Crossing time
using linear regression (with crossing time in seconds
as the continuous outcome) and logistic regression (for Next, we investigated and analysed pedestrian crossing
unsafe crossing behaviours with binary outcomes; for times and other unsafe crossing behaviours among the
instance, sudden stop vs. otherwise). distracted and undistracted (see Table 2). For those vio-
lating the red phases, their crossing tasks were aborted,
and their crossing times, as well as their sudden-stop
3. Results behaviour, were thus not measured. Table 2 presents
the proportion of unsafe crossing behaviour by partici-
3.1. General results
pants in each distracted group. For instance, 8.4% for
All participants were recruited from Taipei Medical Uni- texting (traditional) means that 8.4% of participants
versity, Taiwan, and the University Hospital; and as who engaged in traditional text messaging did not look
many as 1702 participants volunteered to participant in both ways before crossing. The mean crossing time for
the study. Although the participants were naïve to the the undistracted participants was 15.6 s, and distracted
purpose of the study, they may have been reminded by participants took longer to cross the street. Among the
other participants (e.g. classmates or friends) who had distracted participants, those texting using an instant-
already completed the survey. Their crossing behaviours messaging app took the longest to cross the street
may therefore have become more alerted, and if this (20.5 s), and the difference was significant compared
occurred, the true prevalence of unsafe crossing may be with the control group (p < .01).
higher than observed. To prevent the data from such Participants text messaging using an app exhibited the
contamination, all participants, including accompanying lowest head-turning frequency before crossing (average
friends, were asked to indicate whether they were aware frequency: 0.7; p < .05). Those text messaging using an
of the study purpose. Around 2% of the cases were ident- instant-message app were the least likely to look both
ified to have known the study purpose (i.e. they will be ways before crossing (13.7%), and the difference was sig-
observed, and they will receive calls/texts from their nificant compared with the control group (p < .01). Par-
accompanying friends), and they were excluded from ticipants text messaging using an app were the most
the experiment. This gives the final sample size of 1669 disobedient (8.9%; p < .05) and stopped the most while
participants. crossing (5.7%; p < .05). Pedestrian texters were the
Table 1 shows the distribution of various distraction least likely to see the clown, regardless of whether they
types according to independent variables. Data analysis were using an app or the traditional method. Those lis-
of demographic characteristics, handset attributes, and tening to music were the least likely to hear the horn
social conformity was conducted using the χ 2 tests. from the clown (68.8%; p < .01).
Some categories with low counts/expected values were Correlations among the variables were not calculated
combined with other categories, although some cat- as all variables, apart from the variable ‘age’, were categ-
egories with zero count were not included in the chi- orical data. Treating the variable ‘age’ as the continuous
squared analyses. For instance, the category ‘faculty’ or categorical data produced insignificant results when
was combined with the category ‘administrators’, and fitting the regression model. The variable ‘age’ was there-
‘none’ was combined with ‘restricted’ for the chi-squared fore excluded from the analyses.
442 P.-L. CHEN ET AL.

Table 2. Unsafe crossing behaviours by experimental and control groups (N = 1669).


Failure to look both
Averaged Average head- ways before crossing Disobeying the Sudden Failure to see Failure to hear
crossing time (s)a turning frequency (%) signal (%) stop (%)a the clown (%) the horn (%)
Experimental group
Texting 18.8* 0.9* 8.4* 7.2** 5.1 17.5* 21.6
(traditional)
Listening to music 15.4 1.7* 3.2 6.6 0.2* 5.7 68.8**
Talking 17.3* 1.3* 5.0 7.8 0.5 12.1 43.7
(traditional)
Talking using an 17.8* 1.1 7.3** 8.1** 0.3 12.6 45.9*
app
Text messaging 20.5** 0.7* 13.7** 8.9* 5.7* 23.6** 31.5
using an app
Control group 15.6 2.3 1.6 6.7 0 3.1 4.7
a
Only those not violating traffic signals were measured (N = 1543).
*p < .05 compared with control group.
**p < .01 compared with control group.

Table 3 reports the average crossing times, and only 3.3. Crossing behaviours and inattentional
those who were not violating the traffic signals were blindness
measured (N = 1543). As reported in Table 3, partici-
Several binary logit models of unsafe crossing behaviours
pants texting using the traditional way (β=0.37, CI =
were estimated. As reported in Tables 4 and 5, ped-
0.21 to 0.60) or an app (β=0.26, CI = 0.13 to 0.52)
estrians texting using an app were the most likely to
took significantly longer to cross the street than did
stop on the zebra crossing (odds ratio [OR] = 2.98, CI
undistracted participants. Participants listening to
= 1.50 to 5.95), disobey the traffic signals (OR = 2.27,
music crossed in the shortest time (β = −0.20, CI =
CI = 1.36 to 3.80), and not look both ways before cross-
−0.39 to −0.09). Student participants took longer to
ing (OR = 2.59, CI = 1.05 to 2.34), and were less likely to
cross the street than those with other occupations (β
see the clown (OR = 2.39, CI = 1.48 to 3.85).
= 0.27, CI = 0.06 to 0.42). Finally, participants using
Among the distracted groups, listening to music was
phones with screens 5 inches or larger, smartphones,
not associated with any unsafe crossing (OR = 1.09 for
and unlimited mobile Internet crossed more slowly
sudden stops, OR = 1.30 for disobeying the signals, and
than did those using phones with screens smaller
OR = 1.24 for not looking both ways). Participants listing
than 5 inches, feature phones, and restricted or no
to music, however, were the most likely (OR = 3.68, CI =
Internet data, respectively.
1.67 to 5.83) not to hear the horn from the clown.
Student participants were found to be more likely to
perform all three types of unsafe crossing behaviours,
but were less likely to exhibit the two types of inatten-
Table 3. Impact of distraction events and human or phone tional blindness. Female participants were most likely
attributes on time to cross (in seconds) (N = 1543). than male participants to perform all unsafe crossing
p- behaviours, though the difference was only marginally
β SE 95% CI Value
significant. However, female participants were less likely
Distraction events
No distraction Ref. to exhibit the two types of inattentional blindness (OR =
Text messaging (traditional 0.37 0.15 0.21 to 0.60 <.01 0.79 for not noticing the clown and OR = 0.86 for not
texting)
Listening to music −0.20 0.16 −0.39 to −0.09 .41
hearing the horn).
Talking (traditional) 0.19 0.07 0.08 to 0.32 <.01 The probability of a participant disobeying the signal
Talking (using an app) 0.26 0.05 0.13 to 0.52 <.01 increased with the number of pedestrians crossing illeg-
Texting messages (using an 0.86 0.20 0.63 to 0.99 <.01
app) ally from the same side. This result is consistent with pre-
Student (ref. otherwise) 0.27 0.08 0.06 to 0.42 <.01 vious studies (Nasar, Hecht, and Wener 2008) in
Female −0.12 0.08 −0.27 to 0.26 .33
Screen size of 5 inches or larger 0.57 0.12 0.36 to 0.83 <.01 suggesting that that the decision to disobey the signals
(ref. otherwise) is influenced by social forces that favour herd behaviour.
Phone type
Smartphone 0.63 0.10 0.43 to 0.97 <.01 Pedestrians are likely to adopt herd behaviour when
Feature phone Ref. crossing a street; that is, they are more likely to cross if
3G mobile Internet
Unlimited use 0.83 0.25 0.65 to 0.95 <.01
another pedestrian or pedestrians do so. This is possibly
Restricted allowance 0.61 0.16 0.41 to 0.77 .02 because pedestrians might feel safer making a dangerous
None Ref. crossing when doing so with others. Further studies may
BEHAVIOUR & INFORMATION TECHNOLOGY 443

Table 4. Odds of the three unsafe crossing behaviours by distraction activities and human or phone attributes (N = 1669).
Failure to look both ways
Sudden stop Disobeying the signals before crossing
OR (95% CI) p-Value OR (95% CI) p-Value OR (95% CI) p-Value
Distraction type
No distraction Ref. Ref. Ref.
Texting messages (traditional) 1.76 (1.25, 2.48) <.01 1.92 (1.28, 2.85) <.01 2.13 (1.34, 3.40) <.01
Listening to music 1.09 (0.97, 1.23) .13 1.30 (0.86, 1.95) .21 1.24 (0.88, 1.76) .22
Talking (using an app) 1.68 (1.06, 2.67) .03 1.71 (1.06, 2.75) .02 2.02 (1.08, 2.99) <.01
Talking (traditional) 1.57 (1.07, 2.30) .02 1.83 (1.09, 3.07) .02 2.30 (1.13, 3.81) <.01
Texting messages (using an app) 2.98 (1.50, 5.95) <.01 2.27 (1.36, 3.80) <.01 2.59 (1.54, 4.43) <.01
Students (ref. otherwise) 1.60 (1.05, 2.43) .03 2.31 (1.09, 4.89) .03 1.56 (1.05, 2.34) .02
Female (ref. male) 1.30 (0.93, 1.82) .13 1.33 (0.89, 2.00) .16 0.82 (0.62, 1.08) .16
Screen size of 5 inches or larger (ref. otherwise) 1.71 (1.23, 2.39) <.01 2.11 (1.34, 3.35) <.01 2.41 (1.41, 3.89) <.01
3G mobile Internet
Unlimited use 2.65 (2.47, 4.79) <.01 1.71 (1.24, 2.38) <.01 1.89 (1.04, 3.45) .03
Restricted allowance 1.53 (1.19, 1.96) <.01 1.31 (1.01, 1.68) .03 1.33 (1.02, 1.75) .12
None Ref. Ref. Ref.
Number of pedestrians crossing illegally from the participant side 1.09 <.01
r2 0.39 0.34 0.41

investigate whether the herd behaviour can be particu- more likely not to look both ways than were participants
larly prevalent among people walking while using mobile with no mobile Internet access. These participants were
phones. also 1.76 and 2.31 times more likely not to see the
Those using phones with screens 5 inches or larger clown and not hear the horn, respectively, than were par-
exhibited an increased likelihood of engaging in the ticipants with no mobile Internet access.
three unsafe crossing behaviours. Participants with larger
screen size tended to fail to report seeing or hearing the
clown (OR = 1.73 for not noticing the clown; OR = 2.03 4. Discussion
for not hearing the horn). Unlimited access to mobile Studies of the association between pedestrian behaviours
Internet was associated with the three unsafe crossing and phone use (see, for instance, Kahneman 1973) have
behaviours; participants with unlimited Internet access reported that mobile phone use was associated with
were 2.65 times more likely to suddenly stop, 1.71 slower crossing times and less cautious behaviours.
times more likely to disobey the signal, and 1.89 times Our experimental study suggests that, in general, those
engaging in distraction activities such as phone calls
and text messaging (whether by using traditional
Table 5. Odds of the two types of inattentional blindness by phone methods or instant-messaging apps) were more
distraction activities and human or phone attributes (N = 1669). likely than undistracted users to perform unsafe beha-
Failure to see the clown Failure to hear the horn viours (e.g. taking longer to cross, stopping suddenly,
p- p-
OR (95% CI) Value OR (95% CI) Value
disobeying the signals, not looking both ways before
Distraction type
crossing, and looking at traffic less frequently). In
No distraction Ref. Ref. addition, those engaging in talking and texting activities
Text messaging 1.60 (1.20, 2.12) <.01 1.58 (1.02, 2.47) .02 are less likely to see the clown or hear the horn. An
(traditional)
Listening to music 1.13 (0.93, 1.40) .39 3.68 (1.67, 5.83) <.01 expected finding is that, consistent with Thompson
Talking (using an 1.97 (0.74, 5.30) .17 2.36 (1.01, 5.51) .03 et al. (2013), individuals listening to music crossed
app)
Talking 1.57 (0.90, 2.72) .11 2.13 (1.02, 4.49) .03 more quickly than did those engaging in other distrac-
(traditional) tion activities. Distractions other than music likely
Text messaging 2.39 (1.48, 3.85) <.01 2.50 (1.51, 4.16) <.01
(using an app) caused slower crossing times because the visual attention
Students (ref. 0.88 (0.73, 1.06) .18 0.95 (0.91, 1.01) .17 of participants shifted from route planning to the dis-
otherwise)
Female (ref. male) 0.79 (0.69, 0.91) <.01 0.86 (0.74, 0.99) .03
tracting task. In our study, listening to music not only
Screen size of 1.73 (1.12, 2.69) .02 2.03 (1.35, 3.07) <.01 reduced crossing time but also impaired hearing ability,
5 inches or larger with participants who listened to music being less likely
(ref. otherwise)
3G mobile Internet to hear the horn from the clown.
Unlimited use 1.76 (1.30, 2.41) <.01 2.31 (1.41, 3.81) <.01 We specifically investigated the determinants of
Restricted 1.36 (1.02, 1.83) .04 1.60 (1.03, 2.55) .05
allowance unsafe crossing behaviours in the distracted group. The
None Ref. Ref. results suggested that those with unlimited mobile Inter-
r2 0.31 0.35
net access tended not to see the clown and hear the horn.
444 P.-L. CHEN ET AL.

Past laboratory studies (Cohen 1988) have reported that controlled for several influential variables, including
distractions impair pedestrians’ awareness of their sur- mobile screen size and 3G Internet access, that have
roundings. Our results further confirm that mobile not been investigated in past studies. The current
phone users, in particular those with unlimited Internet research has also controlled for whether calls and text
access and those text messaging using an app, tended to messaging were made or transmitted using an instant-
exhibit visual and sound blindness (i.e. were less likely to messaging app or the traditional method. The results
see the clown or hear the horn). showed that app communication was associated with
The present paper also contributes to the safety- the three unsafe crossing behaviours.
research community by concluding that phones screens Our study is limited to the designed content of phone
5 inches or larger and unlimited Internet access are conversation and texting messaging; more engaging or
associated with the three unsafe crossing behaviours. more complicated conversations and text messages
Possible interventions may include educating the public may cause more unsafe crossing behaviours. Finally,
about dangerous crossing behaviours that may arise the present paper analysed pedestrian unsafe crossing
from mobile phone use in general, and from phones behaviours, but it was not our attempt to link these beha-
with large screens and unlimited Internet access in par- viours to accident/injury risks. In addition, our quasi-
ticular. Our result relating to large screen size may experimental study was conducted on the street (where
guide phone manufacturers in their decision-making the speed limit is 20 km/h, and controlled by automatic
process; although handsets with large screens can be signals) connecting the university campus and the hospi-
more eye-catching, the divided attention caused by tal, and in daylight conditions. Therefore, the results may
such screens while driving or walking should not be not be representative of other locales and times.
overlooked.
The rapid increase in the popularity of text messaging,
in particular through an instant-messaging app, suggests
that the risk of distraction and a subsequent accident or
5. Conclusions
injury will increase. Besides education, counter-measures With the rapid development of smartphone technology,
may include engineering and environmental modifi- investigating how text messaging using apps and the use
cations, as well as enforcement efforts. Environmental of other phone features affects unsafe crossing beha-
modifications separating pedestrians from motor traffic viours is crucial. The present study demonstrated that
and promoting conflict-free crossings may be effective text messaging (particularly through instant-messaging
in areas with numerous jaywalking pedestrians. The apps) in a real-world environment increased crossing
enforcement of laws against dangerous walking, includ- time, reduced head-turning frequency before crossing,
ing texting while walking as implemented in Fort Lee, and increased the likelihood of unsafe crossing beha-
New Jersey, may also be considered in Taiwan to reduce viours such as not looking both ways before crossing,
the risk of distractions and accidents. disobeying the signals, and stopping suddenly. In
Notably, although some measures such as disobeying addition, text messaging using apps caused inattentional
the signal, head-turning frequency, and not looking both blindness, with participants who engaged in this activity
ways affect safety, the relevance to safety of outcome being less likely to see the clown or hear the horn played
variables such as crossing speed and stopping behaviours by the clown. The current research also contributes to
is questionable. Although these arguments seem valid, a the growing body of literature on technological distrac-
reduced walking speed, for instance, may leave ped- tions and pedestrian crossing behaviours by demonstrat-
estrians stranded in the street, endangering the safety ing that large phone screen size (5 inches or above) and
of them and other road users. One may also argue that unlimited 3G Internet access impair the safety of ped-
sudden stops are probably not associated with an estrians who text message while walking. Our result
increased accident risk; however, such behaviour may relating to large screen size may provide phone manufac-
delay other pedestrians, reducing their time to finish turers with guidance on their decision-making process;
crossing. The present paper analysed pedestrian unsafe undoubtedly large screens can be more eye-catching,
crossing behaviours, but is not intended to link these but the divided attention caused by such large screens
behaviours to accident or injury risks. Readers are rec- while driving or walking should not be overlooked.
ommended to bear this in mind and that the results
should be interpreted with caution.
Similar to previous observational research, the current
study has strengths as well as limitations. We observed Disclosure statement
numerous individuals in a real-life environment and No potential conflict of interest was reported by the authors.
BEHAVIOUR & INFORMATION TECHNOLOGY 445

Funding Nasar, J., P. Hecht, and R. Wener. 2008. “Mobile Telephones,


Distracted Attention, and Pedestrian Safety.” Accident
The research was funded by the research grants from Ministry Analysis and Prevention 40: 69–75.
of Science and Technology, Taiwan [102-2221-E-038-017- Neider, M., J. McCarley, J. Crowell, H. Kaczmarski, and A.
MY2; 103-2221-E-038-020-MY2]. Kramer. 2010. “Pedestrians, Vehicles, and Cell Phones.”
Accident Analysis and Prevention 42: 589–594.
Newcombe, A. 2012. “Texting while Walking Banned in New
Jersey Town.” American Broadcasting Coporation (ABC)
References News.
Apple Daily. 2014. “A Fine of NTD300 is Proposed.” Accessed Olson, R., R. Hanowski, J. Hickman, and J. Bocanegra. 2009.
December 19, 2014. http://www.appledaily.com.tw/ Driver Distraction in Commercial Vehicle Operations.
realtimenews/article/new/20140504/391453/. Report No. FMCSA-RRR-09-042. Washington, DC:
Byington, K., and D. Schwebel. 2013. “Effects of Mobile Federal Motor Carrier Safety Administration.
Internet use on College Student Pedestrian Injury Risk.” Owens, J., S. McLaughlin, and J. Sudweeks. 2011. “Driver
Accident Analysis and Prevention 51: 78–83. Performance while Text Messaging using Handheld and
Caird, J., K. Johnston, C. Willness, M. Asbridge, and P. Steel. In-vehicle Systems.” Accident Analysis and Prevention 43:
2014. “A Meta-analysis of the Effects of Texting on 939–947.
Driving.” Accident Analysis and Prevention 71: 311–318. Rubinstein, J., D. Meyer, and J. Evans. 2001. “Executive
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Control of Cognitive Processes in Task Switching.”
Sciences. Hillsdale, NJ: Elsevier. Journal of Experimental Psychology Human Perception
Collet, C., A. Guillot, and C. Petit. 2010. “Phoning while Performance 27: 763–797.
Driving I: A Review of Epidemiological, Psychological, Schabrun, M. 2014. “Texting and Walking: Strategies for
Behavioural and Physiological Studies.” Ergonomics 53 (5): Postural Control and Implications for Safety.” PloS One 9
589–601. (1): e84312.
Hancock, P., L. Simmons, L. Hashemi, H. Howarth, and T. Simons, D. 2000. “Attentional Capture and Inattentional
Ranney. 1999. “The Effects of In-vehicle Distraction on Blindness.” Trends in Cognitive Science 4: 147–155.
Driver Response During a Crucial Driving Maneuver.” Strayer, D., F. Drews, and W. Johnston. 2003. “Cell Phone-
Transportation Human Factors 1 (4): 295–309. induced Failures of Visual Attention During Simulated
Horrey, W., and D. Simons. 2007. “Examining Cognitive Driving.” Journal of Experimental Psychology: Applied 9:
Interference and Adaptive Safety Behaviours in Tactical 23–32.
Vehicle Control.” Ergonomics 50 (8): 1340–1350. Thompson, L., R. Rivara, R. Ayyagari, and B. Ebel. 2013.
Hosking, S., K. Young, and M. Regan. 2009. “The Effects of “Impact of Social and Technological Distraction on
Text Messaging on Young Drivers.” Human Factors 51 Pedestrian Crossing Behaviour: An Observational Study.”
(4): 582–592. Injury Prevention 19: 232–237.
Hyman, I., S. Matthew Boss, B. Wise, K. McKenzie, and J. Young, K., C. Rudin-Brown, C. Patten, R. Ceci, and M. Lenne.
Caggiano. 2010. “Did You See the Unicycling Clown: 2014. “Effects of Phone Type on Driving and Eye Glance
Inattentional Blindness while Walking and Talking on a Behaviour while Text-messaging.” Safety Science 68: 47–54.
Cell Phone.” Applied Cognitive Psychology 24: 597–607. Zhou, R., W. Horrey, and R. Yu. 2009. “The Effect of
Hyman, I., B. Sarb, and B. Wise. 2014. “Failure to See Money Conformity Tendency on Pedestrians’ Road-crossing
on a Tree: Inattentional Blindness for Objects that Guided Intentions in China: An Application of the Theory of
Behavior.” Frontiers in Psychology 5 (356): 1–7. Planned Behavior.” Accident Analysis and Prevention 41
Kahneman, D. 1973. Attention and Effort. Englewood Cliffs, (3): 491–497.
NJ: Prentice Hall. Zimbardo, P., and M. Leippe. 1991. The Psychology of Attitude
Lamberg, E., and L. Muratori. 2012. “Cell Phones Change the Change and Social Influence. 3rd ed. New York: McGraw-
Way We Walk.” Gait and Posture 35 (4): 688–690. Hill.
Copyright of Behaviour & Information Technology is the property of Taylor & Francis Ltd
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.

You might also like