Texting and Walking: A Controlled Field Study of Crossing Behaviours and Inattentional Blindness in Taiwan

Considering the impact that texting may have on driving safety, recently concerns have been raised about how texting may affect pedestrian safety. Results from the studies investigating texting and walking indicate that texting creates a significantly greater interference effect on walking than does listening to music or talking on a mobile phone; compared with pedestrians who were using their phones for other purposes (e.g. talking or listening to music), pedestrian texters were less able to maintain walking speed and more likely to pose a threat to public safety by ignoring their surroundings (or ambling across the street) [1, 2]. A laboratory study by Byington and Schwebal [3] examined whether browsing the Internet on a smartphone while crossing a virtual street compromised the safety of young adult pedestrians, and the study reported that distracted pedestrians were found to wait longer to cross the street, missed more safe opportunities to cross, took longer to initiate crossing when a safe gap was available, looked left and right less often, spent more time looking away from the road and were more likely to be hit or almost hit by an oncoming vehicle. Research has suggested that texting while walking affects gait performance (i.e. ability to walk and balance), which may affect the safety of pedestrians: Rubinstein et al. [4] reported that walking and texting on a mobile phone may modify gait performance because of the increased cognitive demand placed on working memory and executive control in performing dual tasks; Lamberg and Muratori [5] recently revealed reduced walking speed and deviation from a straight path while texting messages; and recently Schabrun [6] demonstrated larger decreases in gait velocity and greater lateral deviation in young adults walking and texting compared with those walking and talking on a phone. As Zhou et al. [7] suggested, conformity is the tendency to follow other individuals’ behaviours, actions, values and ideas to avoid potential conflict with others, and literature has reported that the greater the cohesion between a group, the greater the group pressure to conform in risky behaviours [8]; in a pedestrian walking behavioural study conducted in China [7], pedestrians were found to have stronger intentions to cross against a traffic signal when surrounding pedestrians were doing so. Research has also shown that inattentional blindness may occur in naturalistic settings caused by mobile phone conversations while driving and walking [9, 10]: Hyman et al. [10] reported that people talking on a mobile phone while walking failed to see a unicycling clown, and Hyman et al. [11] later pointed out that their participants when walking and texting messages were less likely to show awareness of money in a tree (that serves as an unusual object) beside the pathway on which they were walking; these past studies have attributed inattentional blindness to the possibility that divided attention in a complex environment reduces awareness of objects that are not the focus of attention, and that the objects that people fail to see can be interesting and surprising, but are not directly related to the person’s primary task. Concurrent with other countries that have outlawed mobile phone use (including texting) while driving, Taiwan has banned mobile phone use while driving (including while riding a motorcycle); official statistics on texting-related pedestrian casualties are few, however, with the advances in smartphone features and functions, using phones while driving may remain a frequent practice. Followed by Fort Lee in the U.S. state of New Jersey, where jaywalking pedestrians may be ticketed if caught texting and walking [12], Taiwan is developing a law to ban dangerous walking behaviour such as texting and walking [13]. The current research improves upon Byington and Schwebal [3] who examined the crossing behaviours of pedestrians distracted by mobile phone Internet browsing in a virtual environment, by conducting a field-controlled study; the main purpose of this study is to investigate the effects of phone use (talking, texting and listening to music) on pedestrian street-crossing behaviours, and the present article also examines the relationship between distractions (i.e. phone use) and inattentional blindness.

To be eligible, participants had to be mobile phone users (either feature phones or smartphones) and be familiar with text messaging and making phone calls; they were ineligible if they reported a history of medical illness relating to vision or hearing, or any conditions that restricted their ability to walk, send a text message or place a phone call. Notably, to identify smartphone users, smartphones can have either touchscreens or keypads, but only those using smartphones with touchscreens were included as keypad smartphones are very rare. In total, 1,669 volunteer participants for the controlled field study were recruited from Taipei Medical University, Taiwan, and the University Hospital, with a small incentive (a convenience store voucher worth approximately US$2) offered to volunteers.

All participants were assigned to either the experimental group (distracted individuals) or the control group (those crossing the street without distractions from listening to music, making a phone call, or text messaging), with assignment based on whether an individual was distracted—distracted by listening to music, making a call, or texting (via app or traditional method) for the experimental group, and undistracted when proceeding across the street for the control group; experimental group participants could use smartphones or feature phones (for music, calls, or traditional texting), while control group participants could use either type, and all were naïve to the study’s purpose, only instructed to cross the street (guided from the lab to the street to cross without additional reason), unaware they were video recorded and unaware of incoming texts or calls serving as distractions. For the effect of listening to music, distracted participants assigned as music listeners were required to play preloaded pop music on their devices, with the same song used for all to avoid bias (different music types’ effects on unsafe crossing are beyond this study’s scope). For the effects of texting and making calls, a pilot survey revealed participants mostly ignored texts/calls from strangers (research assistants), so each experimental group member was asked to bring a friend, who was instructed to call or text the participant before crossing; upon arriving at the lab, participants and their friends were isolated and given individual instructions, with CCTV monitoring used by research assistants to guide friends on when to send texts or make calls. If both participant and friend had the same instant-messaging app (e.g., LINE, WhatsApp), texts/calls used the app; otherwise, traditional methods were adopted, with the designed text conversation: “I forgot our tutor’s/supervisor’s phone number. Please tell/text me the number. Thanks!”—participants who ignored the contact and crossed were considered undistracted, while others either texted back or called. Data were collected from May 2012 to April 2015 on weekdays during three time periods: morning peak hours (0700–0900), off-peak time (1200–1400), and afternoon peak hours (1600–1800), with evening observations avoided due to poor visibility that could increase accident risks for participants (Fig. 1).

All participants were recruited from Taipei Medical University, Taiwan, and the University Hospital, with 1,702 volunteers initially participating; though participants were naïve to the study’s purpose, some might have been informed by peers who had completed it, potentially making their crossing behaviours more alert and leading to underobserved unsafe crossing prevalence. To avoid data contamination, all participants and their accompanying friends were asked about their awareness of the study purpose, and around 2% who knew (i.e., they knew they would be observed and receive calls/texts from their friends) were excluded, resulting in a final sample of 1,669 participants—of the 896 in the distracted group, 450 and 446 crossed during red and green traffic phases respectively, while of the 773 in the undistracted group, 390 and 383 faced red and green phases respectively. Next, pedestrian crossing times and other unsafe crossing behaviours in both groups were investigated and analysed (see Table 1); for those violating red lights, their crossing tasks were aborted, so crossing times and sudden-stop behaviours were not measured. Table 1 shows the proportion of unsafe crossing behaviours in each distracted group—for example, 8.4% for traditional texting indicates 8.4% of those using traditional text messaging did not look both ways before crossing. The mean crossing time for undistracted participants was 15.6 seconds, while distracted participants took longer; among the latter, those texting via instant-messaging apps had the longest crossing time (20.5 seconds), a difference significant compared to the control group (p < 0.01). Participants texting via apps also had the lowest head-turning frequency before crossing (average 0.7; p < 0.05), were the least likely to look both ways before crossing (13.7%, p < 0.01), the most likely to disobey traffic signals (8.9%, p < 0.05), and the most likely to stop while crossing (5.7%, p < 0.05). Pedestrian texters, whether using apps or traditional methods, were the least likely to see the clown, while those listening to music were the least likely to hear the clown’s horn (68.8%, p < 0.01). Table 2 reports average crossing times, measured only for participants who did not violate traffic signals (N = 1,543); as shown, participants texting via apps crossed significantly longer than undistracted ones, while those listening to music had the shortest crossing time, and student participants crossed longer than those with other occupations. Additionally, participants using phones with screens 5 inches or larger, smartphones, or with unlimited mobile Internet crossed more slowly than those using smaller-screen phones, feature phones, or with restricted/no Internet data. Several binary logit models of unsafe crossing behaviours were estimated, and as reported in Tables 3 and 4, pedestrians texting via apps were the most likely to stop on zebra crossings (odds ratio [OR] = 2.98, p < 0.01), disobey traffic signals (OR = 2.27, p < 0.01), not look both ways before crossing (OR = 2.59, p < 0.01), and less likely to see the clown (OR = 2.39, p < 0.01). Among distracted groups, listening to music was not associated with any unsafe crossing behaviours (OR = 1.09 for sudden stops, OR = 1.30 for signal disobedience, OR = 1.24 for not looking both ways), but those listening to music were the most likely not to hear the clown’s horn (OR = 3.68). Student participants were more likely to engage in all three types of unsafe crossing behaviours but less likely to exhibit two types of inattentional blindness; female participants were more likely than males to perform all unsafe crossing behaviours (though the difference was only marginally significant) but less likely to show inattentional blindness (OR = 0.79 for not noticing the clown, OR = 0.86 for not hearing the horn). The probability of a participant disobeying traffic signals increased with the number of pedestrians illegally crossing from the same side, consistent with previous studies [14] suggesting signal disobedience is influenced by social forces favouring herd behaviour—pedestrians are more likely to cross if others do so, possibly because they feel safer in a group, and this behaviour is particularly common among those using mobile phones while walking. Participants using phones with screens 5 inches or larger had a higher likelihood of engaging in three unsafe crossing behaviours, and larger screens also impaired visual and auditory abilities (OR = 1.73 for not noticing the clown, OR = 2.03 for not hearing the horn). Unlimited mobile Internet access was associated with three unsafe crossing behaviours: those with unlimited access were 2.65 times more likely to suddenly stop, 1.71 times more likely to disobey signals, 1.89 times more likely not to look both ways, 1.76 times more likely not to see the clown, and 2.31 times more likely not to hear the horn compared to those with no mobile Internet access.

We specifically investigated the determinants of unsafe crossing behaviours in the distracted group; the results suggested that those with unlimited mobile Internet access tended not to see the clown and hear the horn. Past laboratory studies [15] have reported that distractions impair pedestrians’ awareness of their surroundings, and our results further confirm that mobile phone users, in particular those with unlimited Internet access and those text messaging using an app, tended to exhibit visual and sound blindness (i.e., were less likely to see the clown or hear the horn). The present article also contributes to the safety-research community by concluding that phone screens 5 inches or larger or unlimited Internet access are associated with the three unsafe crossing behaviours; possible interventions may include educating the public about dangerous crossing behaviours that may arise from mobile phone use in general and from phones with large screens and unlimited Internet access in particular, and our result relating to large screen size may guide phone manufacturers in their decision-making process—although handsets with large screens can be more eye-catching, the divided attention caused by such screens while driving or walking should not be overlooked. The rapid increase in the popularity of text messaging, in particular through instant-messaging apps, suggests that the risk of distraction and a subsequent accident or injury will increase; besides education, counter-measures may include engineering and environmental modifications, as well as enforcement efforts, such as environmental modifications that separate pedestrians from motor traffic and promote conflict-free crossings (effective in areas with numerous jaywalking pedestrians) and the enforcement of laws against dangerous walking (including texting while walking, as implemented in Fort Lee, New Jersey), which may also be considered in Taiwan to reduce the risk of distractions and accidents. Arguably, the clown may not appear within the focus of attention and may be blocked by vehicles or other pedestrians, so efforts were made to avoid these problems: first, a 185-cm-tall student was recruited to play the clown to avoid being blocked by regular-size cars (though not by larger vehicles such as buses); second, the clown was standing right next to a parking lot where traffic may pose a conflict with crossing pedestrians, making the clown more likely to appear within the visual field of participants. We understand that the position of the clown was critical in determining whether he appeared within the participants’ visual field, but as in well-known studies of inattentional blindness [10, 11, 16], controlling this variable to such an extent was beyond the scope of our experimental design. Notably, although some measures such as disobeying the signal, head-turning frequency and not looking both ways affect safety, the relevance to safety of outcome variables such as crossing speed and stopping behaviours is questionable—while these arguments seem valid, a reduced walking speed, for instance, may leave pedestrians stranded in the street, endangering the safety of them and other road users, and though one may argue that sudden stops are probably not associated with an increased accident risk, such behaviour may delay other pedestrians, reducing their time to finish crossing. The present article analysed pedestrian unsafe crossing behaviours but is not intended to link these behaviours to accident or injury risks, so readers are recommended to bear this in mind and interpret the results with caution. Similar to previous observational research, the current study has strengths as well as limitations: we observed numerous individuals in a real-life environment and controlled for several influential variables, including mobile screen size and 3G Internet access, that have not been investigated in past studies, and the current research has also controlled for whether calls and text messaging were made or transmitted using an instant-messaging app or the traditional method, with results showing that app communication was associated with the three unsafe crossing behaviours.