Analysis of factors affecting drivers’ choice to engage with a mobile phone while driving in Beijing

Transportation Research Part F 37 (2016) 1–9

Contents lists available at ScienceDirect

Transportation Research Part F journal homepage: www.elsevier.com/locate/trf

Analysis of factors affecting drivers’ choice to engage with a mobile phone while driving in Beijing Jing Shi a,⇑, Yao Xiao a, Paul Atchley b a b

Institute of Transportation Engineering, Tsinghua University, Beijing 100084, China Department of Psychology, The University of Kansas, USA

a r t i c l e

i n f o

Article history: Received 13 May 2014 Received in revised form 31 July 2015 Accepted 13 December 2015

Keywords: Mobile phone Driving distraction Risk analysis Structural equation modeling

a b s t r a c t In the current work, we examined reasons that drivers choose to engage with a mobile phone while driving in Beijing. An Internet survey was administered to collect data about talking and texting while driving. Conversations were sorted into different types. Respondents were requested to indicate the frequency of initiating a call or text, perceived risk, perceived importance and emotionality of the call or text. A structural equation model of talking and texting while driving was developed with perceived risk, perceived importance and emotionality of the call as predictors and compared to a similar model with U. S. drivers. Unlike the U.S. data, perceived risk has a significant negative impact on the choice to call or text among drivers in Beijing. Results also show that perceived importance of the call is a major factor affecting the usage of phone while driving. Even though drivers know it is dangerous and illegal, Beijing drivers choose to talk on mobile phones while driving, but they prefer not to text. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Driving distraction, especially mobile phone use while driving, has become common. Mobile phone use while driving diverts attention away from driving and increases the likelihood of traffic accidents (McEvoy & Woodward, 2006). In China, the Standing Committee of the National People’s Congress (2003) has created legislation to restrict the use of mobile phones while driving, both for handheld and hands-free phone use, but mobile phone use while driving still persists. This study aimed to explore why drivers in large urban areas in China use mobile phones while driving. A number of studies have focused on the risks associated with phone use while driving, finding that even hands-free phone use impairs driving performance (Haigney, Taylor, & Westerman, 2000; Horrey & Wickens, 2006; Nasar, Hecht, & Wener, 2008). Treffner and Barrett (2004) studied drivers on a closed circuit driving track to study the influence of hands-free mobile phone use while driving. They showed that a driver’s sensitivity to perceptual information and their perception and awareness of the road conditions were significantly degraded. Hancock, Lesch, and Simmons (2003) examined the effects of phone use on drivers’ actions during a crucial driving maneuver. Drivers were requested to respond to a phone when faced with making a stopping decision while driving a real vehicle around the test track. In the dual-task condition, drivers responded slower to the stop-light and they braked more intensely. Many researchers (Backer-Grøndahl & Sagberg, 2011; Caird, Willness, Steel, & Scialfa, 2008; Holland & Rathod, 2012; Horberry, Anderson, Regan, Triggs, &

⇑ Corresponding author at: Room 311, Department of Civil Engineering, Tsinghua University, Beijing 100084, China. Tel./fax: +86 10 62772300. E-mail address: [email protected] (J. Shi). http://dx.doi.org/10.1016/j.trf.2015.12.003 1369-8478/Ó 2015 Elsevier Ltd. All rights reserved.

2

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

Brown, 2006; Ishigami & Klein, 2009) have indicated that both hand-held and hands-free phone use while driving increases accident risk, and the frequency of using a phone while driving significantly correlates with crash involvement (Gras et al., 2007). Simply using a phone while at a signalized intersection increased accident risk as well (Liu & Lee, 2005). One important question is why drivers continue to engage in this risky behavior. It may be that drivers are unaware of the risk or that mobile phone technology is so compelling that drivers ignore known risks. One study conducted in France, where phoning while driving is legally restricted, tried to identify the profiles of drivers who talk on a phone while driving (Brusque & Alauzet, 2008). They collected samples by interviewing the French drivers by telephone. Results showed that for men, the main explanatory factor of phone use while driving was age, followed by work-related reasons and extensive phone use in daily life. For women, high mileage and intensive phone use were the only two factors. Business drivers made frequent and extensive use of mobile phones while driving (Hislop, 2012). Other studies have examined the effect of attitudes, norms, control factors, and risk perceptions on drivers’ mobile phone use while driving, which are almost based on questionnaire survey. In one such study, researchers utilized the theory of planned behavior (TPB) (Ajzen, 1991) as a predictor for mobile phone use and found that attitudes consistently predicted drivers’ intentions to use a mobile phone while driving and that drivers would answer the call if the caller was a significant other, such as a boyfriend or girlfriend (Walsh, White, Hyde, & Watson, 2008). In general, perceptions of control and perceived risk were not predictors for safer driving intentions. A similar disconnect between perceived risk and phone use while driving has been found among college students for calling (Nelson, Atchley, & Little, 2009) and texting while driving (Atchley, Atwood, & Boulton, 2011). Zhou, Rau, Zhang, and Zhuang (2012) found that intention to answer and perceived behavioral control were consistent predictors of whether drivers engage in compensatory behaviors, such as using hands-free devices or limiting call length. The contents of conversation and driving conditions also influence phone use while driving. Nelson et al. (2009) examined reasons why younger drivers choose to talk on a phone while driving and showed that young drivers still initiated or answered some phone calls even though they perceived the risk of this behavior, because they perceived the calls as important. A study that attempted to examine the relationship between driving conditions and dangerous driving behaviors by simulation found that drivers had more dangerous behaviors when they were in an emotional call than in no call or a mundane call (Dula, Martin, Fox, & Leonard, 2011). Texting is not as common as talking while driving and is the subject of less research. Harrison (2011) found more than 50% of college students in United States texted fairly often while driving. A study of driving simulation indicated that texting decreased the time drivers spent observing the road and increased their workload (Young, Rudin-Brown, Patten, Ceci, & Lenné, 2014). The impairment is worse for older drivers (Rumschlag et al., 2015). Yan, Wong, Li, Sze, and Yan (2015) found impairment for both English and Chinese texting input styles, resulting in increased reaction time and driving speed fluctuations. The research clearly shows that drivers will use phones despite knowing the risk and despite laws that prohibit their use. In order to develop effective safety campaigns, we must understand driver motivations. These motivations may be highly influenced by culture, as well, leading us to want to understand what motivates drivers in China to use phones while driving. Numerous recent studies have suggested that drivers in Beijing may engage in a number of driving practices that are culturally unique (Shi, Bai, Ying, & Atchley, 2010; Shi, Bai, Tao, & Atchley, 2011). Almost all the recent literature investigates drivers in developed countries, despite clear cultural differences that lead to different outcomes for driving safety culture (Atchley, Shi, & Yamamoto, 2014). The current study examines the factors influencing phone use while driving, for both texting and talking, in a representative developing country, China. The focus of this study is to explore how often people talk and text on a mobile phone while driving in Beijing, and the factors that affect their decisions. The current study referred to the method of Nelson et al. (2009) and Atchley et al. (2011), to make the work more directly comparable to an established driving culture (the United States, in this case). In addition, this study explored the influence of different factors between texting and calling while driving, which was not considered in the previous research such as personal characteristics, conversation types and driving conditions. The relationships among the perceived importance of the call, perceived risk, emotionality of the call and the frequency of phone use while driving were also examined. We hypothesized that the perceived importance of the call will encourage the usage of phone while driving, while perceived risk discourages the usage. We also hypothesized that the perceived risk of texting is higher than talking, which may lead to different frequency of the two behaviors. 2. Method 2.1. Participants We conducted questionnaire surveys via the web to obtain the data for this study. A link to a survey entitled ‘‘Survey of Car Drivers’ Distracted Driving Behavior in Beijing” was published on the website http://www.sojump.com/jq/1128826.aspx, and the IP, user name of respondents and answering time were collected to eliminate multiple entries from the same driver. Shi et al. (2010) have shown the results of paper surveys and an online survey have good consistency in a sample of Chinese drivers. An Internet survey can also be cost effective, producing quick responses (Weible & Wallace, 1998) and resulting in less missing data (Stanton, 1998). The participants were restricted to non-commercial drivers. 414 responses were collected

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

3

after removing 74 invalid questionnaires, including samples from outside of the Beijing area (participants whose home address is not in Beijing), duplicate surveys, surveys with unusual answering times (less than 300 s or more than 1800 s), and one participant who did not own a mobile phone. Both questionnaires were removed for participants who completed the survey twice. 2.2. Measures The questionnaire used in this research consisted of four parts. Parts two and three of the questionnaire were designed based on the questionnaires used in Nelson et al. (2009) and Atchley et al. (2011), and the specific items were adjusted to Chinese population. A pilot investigation was implemented to adjust the questions by emailing the questionnaire to a focus group of 30 participants. Statements were modified and common conversation types when driving were collected. Part one measured demographics: gender, age, occupation, monthly income, marital status, educational background, years of driving, car ownership and driving frequency (times/week). In addition, the participants were asked what other distracted behaviors they have when driving. Part two measured self-reported phone use while driving. The participant was asked whether he/she owned a mobile phone and whether he/she used it for texting as well as calling. Then the participants indicated how frequently they read texts, replied to texts, sent texts, answered a phone call, initiated a phone call, and the perceived risk of these actions while driving. The ratings of frequency ranged from 1 to 7: 1 – never, 2 – hardly ever, 3 – occasionally, 4 – sometimes, 5 – often, 6 – frequently, 7 – nearly all the time. To compare the perceived risk of talking and texting, the participants were asked which one is more dangerous. Participants were also asked what type of device they use while driving (hands-free phone with an earpiece/hands-free phone with a speaker/handheld phone) and whether they knew using phone while driving is illegal. Part three asked the same questions about ten different conversation types. The four questions were ‘‘how often you make this type of call while driving”, ‘‘how important it is to make this type of call while driving”, ‘‘how risky it is to make this type of call while driving”, and ‘‘how emotional this type of call tends to make you”. The ten conversation types were based on Nelson et al. (2009) and revised according to interviews in order to make them more suitable for Chinese drivers. The mobile phone conversation types were catching up with friends, business-related activities, talking with significant other, calling out of boredom, calling while driving so it does not have to be done later (multi-tasking), giving or receiving directions, calling to tell your date/appointment your location/arriving time, talking about/making plans, discussing current events or sports, and wanting or needing to know something instantly. The pilot investigation showed that talking on a phone is more common than texting, so the conversation types in this survey were restricted to talking on a phone and participants who did not text could also fill in the questionnaire. Since drivers do not know the content of conversation exactly before answering a call, this questionnaire only asked about initiating a phone call. The options of all the 40 questions ranged from 1 to 7 where 1 is not often, not important, not risky, or not emotional, while 7 is very often, very important, very risky, or very emotional, respectively. Part four consisted of 12 items about calling given different driving conditions. It contains three driving mood conditions: calm, normal and intense mood. The driving mood condition here means the mood of driver before engaging with a phone, while the emotionality of the conversation, mentioned in part three, is how emotional the call tended to make the driver. The calm condition means that driver has control of his emotions, is sober-minded, and positive, while the intense condition means emotional, impulsive, unquiet, and negative. The normal condition is in between these two. Participants were asked to report how likely they were to answer or initiate a conversation through phone while driving in varying driving conditions (calm, normal, or intense), and how long they would maintain that conversation if they either answered or initiated the call. The participants were asked to respond to the items in the 7-point Likert scale where 1 referred to not likely or very short, and 7 referred to very likely or very long. In order to provide guidance for further research, we also asked the participants whether they performed other distracted behaviors when driving, such as snacking, drinking, watching a DVD, listening to the radio, or talking with the passengers. 2.3. Analytical methods Consistent with the approach used in Nelson et al. (2009) and Atchley et al. (2011), structural equation modeling (SEM) was used to analyze the data from the survey to explore the influence of perceived risk and perceived importance on talking and texting on a phone. SEM is an analytic procedure that does an excellent job of accounting for measurement error (unreliability) and missing data, and can test a model overall rather than coefficients individually. It has been used, for example, to model travel behavior (Golob, 2003). Model fit was evaluated using Root Mean Squared Error of Approximation (RMSEA), Chi Square, Non-Normed Fit Index (NNFI) and Comparative Fit Index (CFI). The 10 conversation types were sorted into three parcels to optimize the quality of the measurement in SEM. The parcels were used as predictors in the model instead of individual questions. Aggregate-level data, which means parcels, are higher in reliability and have higher communality relative to item-level data and the indexes of model fit are more acceptable when parcels are used (Little, Cunningham, Shahar, & Widaman, 2002). The conversation types were randomly assigned to the three parcels, which are shown in Table 1. For each conversation type there were four measured components: the frequency of initiating, perceived risk of different conversation types, perceived importance and emotionality of the conversation. Perceived risk of texting, perceived risk of

4

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9 Table 1 Components of parcels. Parcel 1 Calling while driving so it does not have to be done later (multi-tasking) Giving or receiving directions Calling to tell your date your location/arriving time Wanting or needing to know something instantly Parcel 2 Calling out of boredom Talking about/making plans Discussing current events or sports Parcel 3 Catching up with friends Business-related activities Talking with significant other

talking, perceived importance and emotionality are exogenous latent variables, while reported frequency of texting and frequency of talking while driving are endogenous latent variables. In SEM, exogenous variables are factors which are independent from other variables in the causal model, and endogenous variables are factors generated by exogenous variables. Perceived importance and emotionality were created by the three parcels. Perceived risk of texting and perceived risk of talking were separated. Perceived risk of texting was created by indicators for perceived risk of reading, sending and replying a text message while driving. Perceived risk of talking was created by indicators for perceived risk of initiating a call and answering a call. These five indicators for perceived risk were obtained from part two of the questionnaire, in which the participants were asked the perceived risk of these actions while driving, rated from 1 to 7 with 1 being the least risky. The factor for frequency of texting was also created by indicators for frequency of reading, sending and replying a text message while driving, which are asked in the part 2 of questionnaire. Frequency of talking was created by two indicators including frequency of initiating a call and frequency of answering a call. 3. Results 3.1. Descriptive statistics Sociodemographic characteristics of the investigation are shown in Table 2. Of the total of 414 samples, 214 were women and 200 were men. The gender ratio was almost 1.0, better than that of our previous study (Shi et al., 2010), partly due to the increasing number of women drivers in Beijing. From 2005 to 2008, women drivers in Beijing grew by 40.5%, which was much higher than the growth rate of male drivers, which was about 27.2% (Cai, 2010). The age ranged from 18 to 55 and 38.9% of the participants were between 26 and 30 years old. Reported occupations were mostly office workers (59.4%), public officials (14.3%) and students (10.6%), and 1.7% of the participants were professional drivers. In the current sample, 95.6% had a bachelor degree or above. 84.1% of the participants reported talking on a phone while driving at least once a week. The means and standard deviations of frequency of mobile phone use and perceived risk for talking and texting while driving are displayed in Table 3. The mean value of frequency of driving is 3.42, meaning driving 3–4 days per week. Of the 414 samples, 401 participants reported using phone to text in daily life and 248 participants reported that they might text while driving.

Table 2 Sociodemographic characteristics of samples. Gender

Male 48.3%

Female 51.7%

Age

18–20 4.1%

21–25 21.5%

26–30 38.9%

31–35 18.1%

36–40 11.8%

40–45 3.6%

46–50 1.4%

Monthly income (CNY)

<2000 12.6%

2000–4000 23.9%

4000–6000 23.9%

6000–8000 15.7%

8000–10,000 13.0%

10,000–20,000 9.2%

>20,000 1.7%

Marital status

Single 41.5%

Married & no kids 16.0%

Married & has kids 42.5%

Years of driving

51 26.6%

1–3 34.1%

3–6 21.5%

6–9 10.4%

>9 7.5%

Educational background

Primary school N

Junior high school 0.5%

High school 4.1%

Bachelor 72.5%

Master 21.5%

Doctor 1.4%

Driving frequency (days/week)

51 26.3%

2 14.0%

3 12.3%

4 8.9%

5 22.5%

6 8.9%

7 7.0%

5

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

Table 3 Means and standard deviations for frequency and perceived risk of talking or texting while driving (n = 414) (1 = not frequently or not dangerous; 7 = very frequently or very dangerous). Item

a b

Frequency

Frequency Frequency Frequency Frequency Frequency Frequency

Perceived risk

Perceived Perceived Perceived Perceived Perceived Perceived Perceived

of of of of of of

drivinga answering a call while driving initiating a call while driving reading message while driving replying message while driving sending message while driving

risk risk risk risk risk risk risk

of of of of of of of

replying message while driving sending message while driving initiating a call while driving reading message while driving answering a call while driving sending a text versus initiating a call while drivingb replying to a text versus answering a call while drivingb

MEAN

SD

3.42 3.33 2.70 2.55 2.15 1.96

1.99 1.34 1.24 1.21 1.16 1.08

5.32 5.07 4.94 4.86 4.67 5.42 5.45

2.14 2.20 1.57 2.15 1.57 1.93 1.95

Response scale ranges from 1 (driving one day or less per week) to 7 (driving seven days per week). Response scale ranges from 1 (much less dangerous) to 7 (much more dangerous).

Giving or receiving directions, calling to tell your date your location/arriving time and wanting or needing to know something instantly are the three most frequent conversation types. The frequency, perceived importance, risk and emotionality of the different conversation types are listed in Table 5. The emotionality of the call (mean = 4.39, SD = 0.20) and perceived risk of conversation types (mean = 4.92, SD = 0.16) showed no significant differences between the ten conversation types but the perceived importance (mean = 3.64, SD = 0.94) varied along with the conversation types, which indicates that the frequency of phone use about different conversation types is mostly decided by the perceived importance of the conversation. The coefficient of correlation between perceived importance and frequency of initiating a call was 0.916 (p < 0.01), while the coefficient of correlation between perceived risk, emotionality of the call and frequency of initiating were 0.250 (p > 0.05) and 0.053 (p > 0.05), respectively. According to the survey, a number of other distractions also occur while driving. 83.1% of the participants reported talking with the passengers while driving and 82.4% of them reported listening to the radio while driving. 53.5% of the participants report drinking water or juice while driving, and 23.6% snack while driving. Only 3.9% of the participants reported watching DVD while driving. While this number is small, the elevated risk of this behavior makes even this small sample extremely dangerous. 3.2. Structural equation modeling The relationship between personal attributes and frequency of mobile phone use while driving is shown in Table 4. Because not all of the participants use a phone for texting, frequency of talking on a phone was chosen for analysis. The frequency of answering a phone call was partially connected with the caller, because the frequency of answering a call was decided by both the frequency of receiving a call and the probability of picking up the phone. So the frequency of initiating a phone call was used to represent the frequency of mobile phone use in the data analysis in Table 4. Gender, age, monthly

Table 4 Means for frequency of initiating a call while driving of different populations.

a

Gender

Male 2.905

Female 2.514

Age

18–20 2.294

21–25 2.416

26–30 2.882

31–35 2.787

36–40 2.918

40–45 2.000

46–50 2.333

Monthly income (CNY)

<2000 2.231

2000–4000 2.414

4000–6000 2.778

6000–8000 2.908

8000–10,000 2.889

10,000–20,000 3.211

>20,000 3.143

Marital status

Single 2.564

Married & no kids 2.818

Married & has kids 2.795

Years of driving

51 2.282

1–3 2.688

3–6 3.000

6–9 2.953

>9 3.065

Educational background

Primary school N

Junior high school 1.500

High school 2.471

Bachelor 2.700

Master 2.775

Doctor 2.833

Driving frequency (days/week)a

51 2.156

2 2.276

3 2.647

4 2.946

5 3.108

6 3.081

Driving frequency ranges from 1 (one day or less per week) to 7 (7 days per week, i.e. everyday).

7 3.621

6

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

Table 5 Means for four factors of different conversation types. Item

Frequency

Importance

Risk

Emotionality

Catching up with friends Business-related activities Talking with significant other Calling out of boredom Multi-tasking Giving or receiving directions Your location/arriving time Talking about/making plans Current events or sports To know something instantly

2.399 2.543 2.478 1.986 2.597 2.816 2.899 2.036 1.766 2.964

3.181 3.990 4.437 2.478 3.575 4.466 4.089 2.867 2.242 5.121

4.862 5.041 5.111 4.797 4.809 4.836 4.606 4.988 5.065 5.092

4.490 4.797 4.577 4.232 4.261 4.099 4.295 4.283 4.444 4.411

income, marital status, educational background, years of driving and driving frequency (times/week) were included in the cross analysis. A confirmatory factor analysis was conducted to examine the reason for mobile phone use while driving with the exogenous latent constructs of perceived risk of texting, perceived risk of talking, reported emotionality, perceived importance and the endogenous latent constructs of reported frequency of texting while driving and reported frequency of talking while driving. The model considered both texting and talking. 13 respondents who reported not to text on a phone, were removed and 401 samples were analyzed in the model. The results of the model are shown in Fig. 1. The goodness-of-fit indices suggest the model is acceptable (v2 = 193.7; df = 82; RMSEA = 0.058; NNFI = 0.962; CFI = 0.974). Perceived risk of texting and talking showed a significant negative correlation with frequency of texting (b = 0.26, p < 0.05) and talking (b = 0.23, p < 0.05) while driving, respectively. Perceived importance had a positive correlation with frequency of talking while driving (b = 0.21, p < 0.05), and had no significant

Fig. 1. Results of the structural model. Significant beta weights are indicated by asterisks.

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

7

direct correlation with frequency of texting while driving (b = 0.06, p > 0.05). Reported emotionality also had no significant direct correlation with frequency of texting (b = 0.01, p > 0.05) and talking (b = 0.03, p > 0.05). Frequency of texting while driving (b = 0.52, p < 0.05) was significantly correlated to frequency of talking (b = 0.42, p < 0.05). 4. Discussion This research examined using mobile phones while driving among drivers in Beijing. The characteristics of mobile phone use of Chinese drivers has not been thoroughly studied, despite the fact that we know that drivers in Beijing engage in many aberrant driving behaviors (Shi et al., 2011). In the current work, we found that 91.8% of the participants reported talking on a phone while driving at least some of the time, a very high use rate in a country with laws banning the use of mobile phone while driving. Considering 82.1% of the participants knew the regulation that talking on a phone while driving is illegal, this phenomenon becomes even more interesting. Because of the shortage of policeman, on-road enforcement of traffic laws is not powerful enough. The probability of apprehension is relatively low when talking on a phone while driving, which makes drivers feel the risk of penalty is low. What was more surprising is that 59.8% of the participants reported texting while driving at least some of the time and 78.3% of them reported reading text messages while driving. 92% of young drivers in America read texts and 81% of them replied to a text while driving (Atchley et al., 2011). In this study, of the 106 young drivers (age = 18–25), 69.8% of them reported reading texts and 57.5% of them reported replying to a text while driving, which is much lower than American young drivers, though still very high. The frequency of phone use was different among the five use conditions (reading text message, replying text message, sending text message, answering a phone call and initiating a phone call). According to Table 3, the mean score of frequency of answering a call while driving is the highest, and followed by initiating a call, reading a text, replying to a text and initiating a text message. The order of perceived risk is almost the reverse (replying to a text, sending a text, initiating a call, reading a text and answering a call). Drivers consider texting to be much more dangerous than talking. Differences in mobile phone use while driving were related to a variety of demographic data. The data in Table 4 indicates that men use phones more frequently than women while driving, which may be due to the greater frequency of businessrelated phone use. It is consistent with the research by Walsh et al. (2008) which found male drivers and business drivers are more likely to use a phone. The driving task is also more demanding for men (Lesch & Hancock, 2004), which may make this behavior especially risky. The effect of marital status showed that married drivers, who might be more responsible for the family and children, used the phone more frequently while driving. The responsibility of marriage does not lead to safer driving behavior, but instead may lead to additional reasons to call, such as coordinating schedules or providing status updates. Age and years of driving also influence phone use. As age increases, the frequency of mobile phone use while driving changes, become higher and then lower, which is shown in Table 4. Drivers between 26 and 40 had the highest frequency of phone use. However, older drivers in this survey drove the most, so increasing drive time is probably not the main factor underlying this effect. Different from previous study which indicated that young drivers (aged 16–24) talked on a phone most frequently while driving (Cramer, Mayer, & Ryan, 2007) in America, this study found that the age with highest frequency of using phone ranges from 26 to 40 in China. The increase in phone use while driving in the 26–40 year-old group may instead reflect an increased familiarity with mobile technology and greater daily use for personal and business needs. This conclusion is consistent with the fact that monthly income, educational background and driving frequency, also have a positive relationship with frequency of phone use while driving. High income and education level, which are strongly related, result in increased phone use while driving, probably reflecting business-related calls. In the SEM model, data suggests that perceived risk and importance are significant predictors for how often a driver will use a phone while driving. The hypothesis was verified that perceived risk has negative relationship with frequency of phone use ( 0.26 for texting and 0.23 for talking), while the perceived importance of the conversation had a positive relationship with them (0.06 for texting and 0.21 for talking). However, there is a big difference between texting and talking. The directional beta parameters from perceived risk to texting and talking are both significant (p < 0.01), but only the directional beta parameters from perceived importance to talking are significant (p < 0.01). The perceived risk outweighs the perceived importance on texting while driving (0.26 > 0.06). This indicates that even the conversation is very important; the driver will prefer to talk than text. And the main factor for a driver who often texts while driving is his ignorance of risk, not the importance of the communication. The result of SEM shows the perceived risk and perceived importance can explain the motivation of mobile phone use while driving. What is unique in the current study, and not seen in the sample examined by Nelson et al. (2009), is that perceived risk had a stronger effect on the likelihood to initiate a call than the perceived importance of the call. In Nelson et al., perceived risk was a weak negative predictor of initiating a call while perceived importance was a strong positive predictor. In the current work, we found that perceived risk was a stronger predictor of decreasing phone use, similar to other studies (Hallett, Lambert, & Regan, 2011). This could be due to two potential differences. First, Nelson et al. surveyed younger drivers so the effect could be due to a greater age range in the current sample. Younger drivers may be more prone to risky driving behavior (though this proposition is not without controversy, see McKnight & McKnight, 2003) or younger drivers may feel more comfortable with their ability to drive and use a phone. It may also be an effect that is unique to drivers in Beijing, with its very high traffic density and riskier road system. While China has made significant gains in road safety over the last 20 years, it currently has about seven times as many fatalities

8

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9 Table 6 Phone use while driving in different driving conditions. Driving condition

Normal

Calm

Intense

How How How How

3.384 2.348 2.261 2.191

3.459 2.751 2.524 2.553

2.295 1.841 1.877 1.867

likely of answering long if answering likely of initiating long if initiating

per million km driven than the United States (World Health Organization, 2013). Beijing drivers may be trying to minimize risk and therefore they may be more sensitive to factors that increase risk, such as initiating a call while driving. Consistent with this idea was the finding that the frequency of texting is mainly influenced by the risk and has little correlation with perceived importance, while the frequency of talking is affected by both perceived risk and importance. Drivers consider texting much more dangerous, and this elevated sense of risk has an impact on their behavior. This shows that drivers in Beijing are weighing the relative perceived riskiness of these distracting behaviors given the riskiness of driving in a dense, and often chaotic, driving environment. In work by Atchley et al. (2011), perceived risk had a weak influence on texting behavior, but that might be because drivers surveyed drove in a less urbanized environment, and thus perceived the risk of driving to be less. This distinction warrants further study. Also consistent with the view that drivers in Beijing try to limit risk are data showing the driving conditions also impact the decision to call while driving. The mean scores for how likely drivers were to answer or initiate a mobile conversation and how long they would maintain that conversation while driving in varying driving conditions (calm, normal, or intense driving conditions) are shown in Table 6. The drivers are slightly more likely to initiate or answer a call while driving in calm or normal conditions than in intense conditions. These data are consistent with the notion that drivers in Beijing attempt to limit risk while driving, by reducing calls during intense traffic situations. How well they are able to do this for important calls is something that should be further studied. There are also some limitations in the current study. The first limitation is the sampling procedure. The investigation was conducted by internet and did not included people who do not use the internet regularly. Although the samples were randomly selected, the participants might still represent a limited sampling frame. Compared with the huge population of Beijing, the samples of this study are relatively small. The second limitation is due to the design of this research. In the model, the perceived risk and importance were considered to influence drivers’ behaviors, while the role of social norms were not examined which was also a variable to affect driver behaviors, according to the theory of planned behavior. Further work will need to establish a more comprehensive model to figure out why drivers engaged with a mobile phone while driving. 5. Conclusion In this study, we found that most drivers in Beijing knew that using phone while driving was illegal and understood the risk of talking on a phone, but they still chose to phone if they perceived the call to be important. Drivers showed different behaviors for texting and talking, and they considered risk more when texting, and more about perceived importance when talking. There was a large correlation between perceived importance of a call and frequency of initiating a call, indicating drivers are not solely considering risk when they make a choice to drive distracted. There is evidence that drivers in Beijing try to mitigate risk by calling less frequently during heavy traffic, suggesting that enhancing safety awareness of drivers by strengthening traffic safety education will play an important role in preventing crashes. However, there is also reason for caution in the high rate of phone use while driving and extreme distracted driving behaviors such as watching a DVD while driving, in a small subset of the sample. Further understanding how to reach these drivers and get them to drive safely will be critical as China seeks to further improve its road safety. Acknowledgement This study is financially supported by the National Natural Science Foundation of China (No. 51578319) and PSA PEUGEOT CITROEN (China) Automotive Trade Co., Ltd. References Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211. Atchley, P., Atwood, S., & Boulton, A. (2011). The choice to text and drive in younger drivers: Behavior may shape attitude. Accident Analysis and Prevention, 43, 134–142. Atchley, P., Shi, J., & Yamamoto, T. (2014). Cultural foundations of safety culture: A comparison of traffic safety culture in China, Japan and the United States. Transportation Research Part F: Traffic Psychology and Behavior, 26, 317–325. Backer-Grøndahl, A., & Sagberg, F. (2011). Driving and telephoning: Relative accident risk when using hand-held and hands-free mobile phones. Safety Science, 49(2), 324–330. Brusque, C., & Alauzet, A. (2008). Analysis of the individual factors affecting mobile phone use while driving in France: Socio-demographic characteristics, car and phone use in professional and private contexts. Accident Analysis & Prevention, 40(1), 35–44. Cai, N. (2010). The analysis of factors influencing female drivers’ traffic accidents. Beijing University of Technology. Masteral dissertation [in Chinese].

J. Shi et al. / Transportation Research Part F 37 (2016) 1–9

9

Caird, J. K., Willness, C. R., Steel, P., & Scialfa, C. (2008). A meta-analysis of the effects of cell phones on driver performance. Accident Analysis & Prevention, 40 (4), 1282–1293. Cramer, S., Mayer, J., & Ryan, S. (2007). College students use cell phones while driving more frequently than found in government study. Journal of American College Health, 56(2), 181–184. Dula, C. S., Martin, B. A., Fox, R. T., & Leonard, R. L. (2011). Differing types of cellular phone conversations and dangerous driving. Accident Analysis & Prevention, 43(1), 187. Golob, T. F. (2003). Structural equation modeling for travel behavior research. Transportation Research Part B: Methodological, 37(1), 1–25. Gras, M. E., Cunill, M., Sullman, M. J., Planes, M., Aymerich, M., & Font-Mayolas, S. (2007). Mobile phone use while driving in a sample of Spanish university workers. Accident Analysis & Prevention, 39(2), 347–355. Haigney, D. E., Taylor, R. G., & Westerman, S. J. (2000). Concurrent mobile (cellular) phone use and driving performance: Task demand characteristics and compensatory processes. Transportation Research Part F: Traffic Psychology and Behaviour, 3(3), 113–121. Hallett, C., Lambert, A., & Regan, M. A. (2011). Cell phone conversing while driving in New Zealand: Prevalence, risk perception and legislation. Accident Analysis & Prevention, 43(3), 862–869. Hancock, P. A., Lesch, M., & Simmons, L. (2003). The distraction effects of phone use during a crucial driving maneuver. Accident Analysis & Prevention, 35(4), 501–514. Harrison, M. A. (2011). College students’ prevalence and perceptions of text messaging while driving. Accident Analysis & Prevention, 43(4), 1516–1520. Hislop, D. (2012). Hanging on the telephone: Mobile phone use patterns among UK-based business travelers on work-related journeys. Transportation Research Part F: Traffic Psychology and Behaviour, 15(2), 101–110. Holland, C., & Rathod, V. (2012). Influence of personal mobile phone ringing and usual intention to answer on driver error. Accident Analysis & Prevention, 50, 793–800. 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 & Prevention, 38(1), 185–191. Horrey, W. J., & Wickens, C. D. (2006). Examining the impact of cell phone conversations on driving using meta-analytic techniques. Human Factors, 48(1), 196–205. Ishigami, Y., & Klein, R. M. (2009). Is a hands-free phone safer than a handheld phone? Journal of Safety Research, 40(2), 157–164. Lesch, M. F., & Hancock, P. A. (2004). Driving performance during concurrent cell-phone use: Are drivers aware of their performance decrements? Accident Analysis & Prevention, 36(3), 471–480. Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or not to parcel: Exploring the question, weighing the merits. Structural Equation Modeling, 9(2), 151–173. Liu, B.-S., & Lee, Y.-H. (2005). Effects of car-phone use and aggressive disposition during critical driving maneuvers. Transportation Research Part F: Traffic Psychology and Behaviour, 8(4), 369–382. McEvoy, S. P., & 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), 630–634. McKnight, A. J., & McKnight, A. S. (2003). Young novice drivers: Careless or clueless? Accident Analysis & Prevention, 35(6), 921–925. Nasar, J., Hecht, P., & Wener, R. (2008). Mobile telephones, distracted attention, and pedestrian safety. Accident Analysis and Prevention, 40, 69–75. Nelson, E., Atchley, P., & Little, T. (2009). The effects of perception of risk and importance of answering and initiating a cellular phone call while driving. Accident Analysis & Prevention, 41(3), 438–444. Rumschlag, G., Palumbo, T., Martin, A., Head, D., George, R., & Commissaris, R. L. (2015). The effects of texting on driving performance in a driving simulator: The influence of driver age. Accident Analysis & Prevention, 74, 145–149. Shi, J., Bai, Y., Tao, L., & Atchley, P. (2011). A model of Beijing drivers’ scrambling behaviors. Accident Analysis & Prevention, 43(4), 1540–1546. Shi, J., Bai, Y., Ying, X., & Atchley, P. (2010). Aberrant driving behaviors: A study of drivers in Beijing. Accident Analysis & Prevention, 42(4), 1031–1040. Stanton, J. M. (1998). An empirical assessment of data collection using the Internet. Personnel Psychology, 51(3), 709–725. Treffner, P. J., & Barrett, R. (2004). Hands-free mobile phone speech while driving degrades coordination and control. Transportation Research Part F: Traffic Psychology and Behaviour, 7(4), 229–246. Walsh, S., White, K., Hyde, M., & Watson, B. (2008). Dialling and driving: Factors influencing intentions to use a mobile phone while driving. Accident Analysis & Prevention, 40(6), 1893–1900. Weible, R., & Wallace, J. (1998). Cyber research: The impact of the Internet on data collection. Marking Research, 10(3), 19–25. World Health Organization (2013). Global status report on road safety 2013. Geneva: WHO. Yan, W., Wong, S. C., Li, Y. C., Sze, N. N., & Yan, X. (2015). Young driver distraction by text messaging: A comparison of the effects of reading and typing text messages in Chinese versus English. Transportation Research Part F: Traffic Psychology and Behaviour, 31, 87–98. Young, K. L., Rudin-Brown, C. M., Patten, C., Ceci, R., & Lenné, M. G. (2014). Effects of phone type on driving and eye glance behaviour while text-messaging. Safety Science, 47–54. Zhou, R., Rau, P.-L. P., Zhang, W., & Zhuang, D. (2012). Mobile phone use while driving: Predicting drivers’ answering intentions and compensatory decisions. Safety Science, 50(1), 138–149.