Accident Analysis and Prevention 75 (2015) 43–54
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
The road user behaviour of school students in Iran Amir Reza Nabipour a , Nouzar Nakhaee b , Narges Khanjani a, *, Hossein Zirak Moradlou c , Mark J.M. Sullman d a
Department of Biostatistics and Epidemiology, Faculty of Public Health, Kerman University of Medical Sciences, Haft Bagh Alavi Blvd., Kerman, Iran Neuroscience Research Center, Kerman University of Medical Sciences, Kerman, Iran c Consultation Center, Tehran Education and Training Organization, Tehran, Iran d Driving Research Group, Cranfield University, UK b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 29 March 2014 Received in revised form 28 October 2014 Accepted 10 November 2014 Available online xxx
The present study developed a Persian version of the Adolescent Road User Behaviour Questionnaire (ARBQ) and investigated the psychometric properties of the scale in a sample of school students in the province of Tehran (Iran). In total 1111 adolescents completed the Persian version of the ARBQ. Exploratory factor analysis, using the shortened 21-item version of the scale revealed the presence of three reliable factors which were also supported using confirmatory factor analysis. According to this research, engagement in dangerous playing in the road was significantly higher among males, residents of large urban areas, students from private schools, students in the south of Tehran, those who reported relatives or friends had been killed in a road crash and those with a personal history of road accidents. Moreover, older adolescents, those who reported relatives or friends having been killed in a road crash and those with a traffic accident history reported higher involvement in unsafe crossing behaviour. Females, older adolescents, residents of small urban areas, students from schools in small urban areas and those with an accident history also reported less frequent engagement in planned protective behaviours. This study confirms that the ARBQ is a useful framework for investigating adolescents’ on-road behaviours in Iran. This research also showed that adolescents put themselves at risk by engaging in hazardous behaviours. As is the case in most countries, this study revealed the need for interventions, such as education and enforcement to improve the on-road safety culture amongst Iranian adolescents. ã 2014 Elsevier Ltd. All rights reserved.
Keywords: Psychometric assessment Adolescents behaviour Pedestrians ARBQ Road safety
1. Introduction Road traffic accidents pose a major public health problem by causing a large number of injuries, disabilities and fatalities, especially in the low and medium-income countries (Peden et al., 2004; Olukoga et al., 2011; Olukoga, 2003). This is also the case for Iran, which in 2010 recorded a total of 414,161 injuries and 23,249 deaths. The number of fatalities translates into a rate of 12.4/10,000 registered vehicles, which is substantially higher than in most developed countries, such as New Zealand (1.2/10,000 registered vehicles), Austria (1.0/10,000 registered vehicles), the USA (1.3/10,000 registered vehicles) and is more than twenty times higher than the 0.6/10,000 registered vehicles reported by the UK (NZTA, 2011; Bahadorimonfared et al., 2013). Furthermore, the Iranian Forensic Medicine Organization (2012)
* Corresponding author. Tel.: +98 34 3132 5102. E-mail address: [email protected] (N. Khanjani). http://dx.doi.org/10.1016/j.aap.2014.11.009 0001-4575/ ã 2014 Elsevier Ltd. All rights reserved.
reported that from 2006–2011 a total of 120,070 people died on Iranian roads and of these 7565 (6.3%) were adolescents aged between 11 and 17 years old. A number of studies have reported that adolescents have an inflated risk of being killed or injured on the roads (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). One of the reasons for this inflated risk is that adolescents carry out a variety of unsafe and potentially risky on-road behaviours which increases the risk of being killed or injured on the road (West et al., 1998; Poudel-Tandukar et al., 2006; Campbell and Keegan, 2000; Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). One important step for improving the safety of Iranian adolescents is to understand the behaviours that put them at greater risk of being killed or injured on the road. Although there is currently no widely agreed upon framework for investigating the on-road behaviours of adolescents, one which has been utilised in several countries across the world is the Adolescent Road User Behaviour Questionnaire (ARBQ; Elliott and Baughan, 2004). Using
44
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
data from adolescent school students in the UK, Elliott and Baughan investigated the psychometric properties of their 43-item ARBQ and found three factors best described their data. The first factor consisted of items relating to crossing the road in an unsafe manner (e.g. run across a road without looking because you are in a hurry), which they called “unsafe crossing behaviour”. The second factor, “dangerous playing in the road” consisted of items involving playing on the road (e.g. hold on to a moving vehicle when riding a bike). In contrast the third factor “planned protective behaviours” was comprised of safety behaviours that reduce the chance of being killed or injured on the road (e.g. wear bright or reflective clothing when out on foot in the dark). The items of the ARBQ measure not only different types of pedestrian behaviours but also a small number involving: bicycles, skateboards and rollerblades (e.g. use lights on your bike when it is dark). Elliott and Baughan also developed a short 21-item version of the scale, which reflected the same three factors and was developed for use in conjunction with other psychometric scales. The psychometric properties of the ARBQ have also been investigated in New Zealand (Sullman and Mann, 2009), Spain (Sullman et al., 2011), Belgium (Sullman et al., 2012) and France (Abou et al., 2008). These studies all supported the presence of the three factors and the three factors were also supported for the 21item version of the scale in Spain and Belgium using confirmatory factor analysis. However, these findings contrast with the New Zealand research which produced a 19-item version of the scale using exploratory factor analysis (Sullman and Mann, 2009). Although there were considerable differences between the four studies which have used the ARBQ, in terms of different: countries, populations, cultures, ethnic groups, languages, traffic conditions and rules, road traffic environments and population densities, there has been a surprising number of consistent findings. For instance, in terms of gender, in all four studies male students reported more frequent engagement in playing on the road than female adolescents (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). Although, the New Zealand study did not show statistically significant differences between gender and unsafe crossing behaviour (Sullman and Mann, 2009), the other three studies all found that males were more likely to be involved in these types of potentially risky behaviours (Sullman et al., 2011, 2012). However, none of the previous studies found a significant relationship between planned protective behaviour and gender. There were also similarities in the relationships the different ARBQ factors had with age. For example, the English and Spanish studies found that older adolescents reported more unsafe crossing behaviour, while also engaging less often in planned protective behaviour and dangerous playing in the road (Elliott and Baughan, 2004; Sullman et al., 2011). In Belgium, adolescents who were 11–12 years of age reported a lower frequency of unsafe road crossing behaviour, than those aged 13–14 years and greater than 15 years of age (Sullman et al., 2012). Finally, the Belgian research also found that younger respondents reported more planned protective behaviour than older adolescents (Sullman et al., 2012), as was also the case in the UK and Spanish studies (Sullman et al., 2011, 2012). However, as the New Zealand research used very different age categories it was not possible to compare their results with the other ARBQ studies (Sullman and Mann, 2009). As well as age and gender, the ARBQ factors have been found to be related to an adolescent’s place of residence. Elliott and Baughan (2004) found that students from large urban areas reported engaging more often in unsafe road crossing behaviours than those from rural and small urban areas. Furthermore, adolescents living in large urban areas reported less frequently engaging in dangerous playing on the roads than adolescents from rural areas, while rural residents reported more planned protective behaviours than those
from small and large urban areas. However, in contrast to the UK research there were no statistically significant differences between living areas and the ARBQ factors in New Zealand, Belgium or Spain (Sullman and Mann, 2009; Sullman et al., 2011). As would be expected, given the fact that the ARBQ investigates behaviours thought to be important for road safety, previous research has found that adolescents who self-reported previous involvement in a road crash also reported more frequently engaging in the two types of potentially risky behaviours and also, unexpectedly, planned protective behaviours (Sullman et al., 2012). Also surprising was the fact that thus far only one study has investigated the relationship the ARBQ factors have with prior crash involvement. As there is only one study, this relationship may be an anomaly or confined solely to Belgium. Therefore, it is important that research be conducted to test this relationship in a novel sample and preferably in a novel country and culture. Another important gap in the literature is that all previous research using the ARBQ has been conducted in developed countries. This is a particularly important limitation as the size of the road safety problem, in terms of injuries and fatalities, is considerably larger in less developed countries. Therefore, the present study investigated the psychometric properties of the ARBQ in Iran, a developing country with a non-European culture and background. The replication of these findings is a critical step to confirm the suitability of the ARBQ as a framework to measure road user behaviours in this at-risk age group. More specifically, the present study investigated the factor structure of the ARBQ in a large sample of Iranian school students using both exploratory and confirmatory factor analysis for the long and short versions of the ARBQ. The study also investigated the relationships the resultant factors had with age, gender, place of residence, along with the individuals' accident history and those of their friends and family. 2. Methods 2.1. Materials The ARBQ (Adolescent Road-User Behaviour Questionnaire) is a self-report tool for investigating how often participants engage in 43 different road-user behaviours. For example, “How often do you not look because you can't hear any traffic coming?” Responses are made on a five point Likert Scale (1 = never to 5 = very often). Previous studies have found that the questionnaire is measuring three latent variables: “unsafe road crossing behaviour”, “dangerous playing on the road” and “planned protective behaviour” (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). These three factors had internal reliability scores of .89, .85 and .76, respectively. The questionnaire also asked about the participants’ demographic and descriptive details, including: age, gender, nationality, living area, mother’s education level, father’s education level, school type (public or private), school grade, location of the school, along with whether they (or their friends and/or family) had been involved in a road crash. 2.2. Setting and participants The present cross sectional study was carried out in secondary schools in the province of Tehran, Iran. The sample was comprised of 1200 students from large and small urban areas which were randomly selected. Tehran city was selected as the large urban area and Pishva was selected as the small urban area within the same province. In the 2011 Nationwide Iranian Census Tehran city had a population of 8,244,535 while Pishva had a population of 75,454 (Iranian Statistics Center, 2011).
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
From Tehran’s 22 municipal districts, two municipal districts were randomly selected, one from the north (high socioeconomic area) and the other from the south (low socioeconomic area). Four public and four private schools (8 schools) were randomly selected from each district. Half of these schools were for boys’ and the other half were girls’ schools. In the town of Pishva four public secondary schools (two boys’ and two girls’ schools) were randomly selected. In total, 20 schools were included in the study, half of which were junior high schools (7th, 8th, 9th grade) and the other half were high schools (10th, 11th, 12th grade). All random
45
selection took place the same way for all aspects of the study. This was to allocate each province, school and participant a number and then to use SPSS to generate the required quantity of random numbers. Once the schools had been selected, permission was sought from the authorities at each school before starting the study. From each school, 120 students from grade 7 to 12 and between 13 and 18 years old were randomly chosen. All students consented to participate in the survey. Participants were guaranteed anonymity and confidentiality of their answers. All students
Table 1 The characteristics of the students participating in this study. Variable
Frequency
percentage
Gender Males Females
557 554
50.1 49.9
Age 13–14 15–18
378 733
34 66
Grade 7 8 9 10 11 12
173 204 188 193 185 168
15.6 18.4 16.9 17.4 16.7 15.1
1043 68
93.9 6.1
Living area Large urban area Small urban area Rural area
840 215 56
75.6 19.4 5
Mother’s education level Illiterate Elementary school (grade 1–5) Guidance school (grade 6–8) High school (grade 9–12) Diploma Academic level
51 143 139 63 384 331
4.6 12.9 12.5 5.7 34.6 29.8
Father’s education level Illiterate Elementary school Guidance school High school Diploma Academic level
45 115 138 85 353 375
4.1 10.4 12.4 7.7 31.8 33.8
Relatives or friends killed ina road crash Yes No
541 570
48.7 51.3
Traffic accident history Yes No
385 726
34.7 65.3
Family motorized vehicle Yes No
882 229
79.4 20.6
Kind of school Public Private
679 432
61.1 38.9
Location of school (according to socioeconomic status) The north of Tehran (large urban, high socioeconomic status) The south of Tehran (large urban, low socioeconomic status) Small urban
412 458 241
37.1 41.2 21.7
Nationality Iranian Afghan
46
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
2.3. Cross cultural adaptation
took 25–30 min to answer the Persian version of ARBQ. Although all 1200 randomly selected students agreed to participate in this project, 89 students did not complete the survey, leaving a total of 1111 participants (response rate of 92.6%). Of the 1111 students who answered the questionnaire, 557 (50.1%) were male and 554 (49.9%) female, with an average age of 15.45 years old (SD = 1.69, range 13–18 years old). The nationality of these students was mainly Iranian and most lived in Tehran city (Table 1). The adolescence period is generally divided into three periods, including: early (aged 11–14), middle (aged 15–18) and late (aged 18–21) periods (Lerner and Laurence, 1999). In the current study, the number of participants in ages 13–14 and ages 15–18 were 378 (34%) and 733 (66%), respectively. Participants were also asked whether they had been involved in an accident as a pedestrian or while riding a bicycle. Over a third (34.7%, 385) reported that they had been involved in an accident on the road. The characteristics of the participants can be seen in Table 1.
2.3.1. Translation First, forward translations were carried out independently by two native translators. These translations were compared and discussed by researchers and the draft Persian version was prepared. Then, two bilingual native English translators independently translated the Persian version back into English. The translators were not aware of the content of the initial English version. Comparisons were then made between the backward translations and the original version, but no major differences were identified. The draft Persian-language version of the ARBQ was developed by consensus between the translators and researchers. At this point, as drinking alcohol is religiously prohibited for Muslims there searchers made the decision to exclude the item “cross less than an hour after drinking alcohol”. 2.3.2. Cognitive testing The draft Persian version of the questionnaire was then tested on 12 students with an age range of 13–18 years old. An interview
Table 2 ARBQ items means and standard deviation of Iran, Belgium, New Zealand, UK and Spain. No. Item (How often do you. .)
Iran
14 5 17 12 19 16 7 2 20 6 34 1 42 36 41 33 3 11 10 9 18 4 32 35 26 30 23 29 31 27 28 25 24 22 21 –
Look both ways before crossing Cross at a place that is well lit when it is dark Check to make sure traffic has stopped before using a pedestrian crossing See a small gap in traffic and “go for it” Keep looking and listening until you get all the way across the road Walk in single file on roads without pavements Get part way across the road and then have to run the rest of the way to avoid traffic Have to stop quickly or turn back to avoid traffic Cross from behind a stationary vehicle Cross between parked cars when there is a safer place to cross nearby Think it is OK to cross safely, but a car is coming faster than you thought Not bother walking to a nearby crossing to cross the road Make traffic slow down or stop to let you cross Cross without waiting for the ‘green man’ Forget to look properly because you are talking to friends who are with you Not notice a car pulling out (say from a driveway) and walk in front of it Cross when you cannot see both ways very well (like on a bend or top of hill) Wear bright or reflective clothing when riding a bike in the dark Forget to look properly because you are thinking about something else Walk facing the traffic when on roads without pavements Use lights on your bike when it is dark Use a lollipop man/lady where there is one available Wear reflective clothing when out on foot in the dark Use a mobile phone and forget to look properly Walking on the road rather than on the pavement Climb over barriers or railings that separate the road from the pavement Not look because you cannot hear any traffic around Run across a road without looking because you are in a hurry Cross whether traffic is coming or not, thinking the traffic should stop for you Wear reflective clothing when crossing the road Wear a cycle helmet when riding a bike Ride on a skateboard (or roller-skates/roller-blades) on the road Not notice an approaching car when playing games in the road Deliberately run across the road without looking, for a dare Hang around in the road talking to friends Run around in a road (e.g. when playing foot ballor bull dog) Ride out into the road on a skateboard without thinking to check for traffic Run into the road to get a ball, without checking for traffic. Hold on to a moving vehicle when riding a bike Hold on to a moving vehicle when riding a skateboard/roller-skates/rollerblades. Play “chicken” by deliberately running out in front of traffic Play ‘chicken’ by lying down in the road and waiting for cars to come along Cross less than an hour after drinking alcohol
R
SD
SD
R
M
R
M
SD
1.13 1 4.17 .95 1 4.17 1.35 5 3.36 1.06 7 3.13 1.34 2 3.87 1.05 2 3.73 1.36 20 2.56 1.13 6 3.14 1.4 3 3.55 1.14 5 3.24 1.41 10 2.86 1.29 20 2.40 1.38 11 2.74 1.04 9 3.01
.95 1.08 1.15 1.08 1.17 1.16 1.04
1 3 2 8 4 11 6
4.08 3.32 3.46 2.69 3.26 2.54 2.82
1.07 1.16 1.30 1.30 1.26 1.38 1.14
1 3 2 13 4 10 8
4.07 3.48 3.76 2.73 3.34 2.81 2.88
1.08 1.18 1.15 1.13 1.24 1.41 1.04
8 9 10 11 12 13 14 15 16 17
2.87 2.77 2.68 2.59 2.58 2.57 2.46 2.45 2.43 2.38
1.31 30 2.06 .91 1.27 14 2.64 1.00 1.30 12 2.70 .99 1.17 26 2.16 .94 1.36 13 2.66 1.05 1.36 9 2.85 1.19 1.4 21 2.50 1.23 1.35 19 2.57 1.16 1.11 28 2.13 .90 1.29 16 2.59 1.01
23 12 11 21 8 24 18 13 22 16
2.23 2.83 2.88 2.38 3.10 2.22 2.42 2.80 2.30 2.45
.94 19 2.34 1.05 16 2.35 1.03 9 2.66 .92 18 2.34 1.13 7 2.72 1.05 15 2.43 1.23 17 2.35 1.06 10 2.65 1.03 24 2.20 .94 13 2.49
1.07 1.17 1.16 1.12 1.22 1.24 1.17 1.15 1.03 1.10
22 11 7 24 9 14 12 5 20 23
2.27 2.79 2.92 2.17 2.85 2.66 2.79 3.05 2.36 2.25
1.01 1.12 1.07 .98 1.14 1.14 1.12 1.12 1.09 .96
18 19 20 21 22 23 24 25 26 27 28 29
2.35 2.33 2.32 2.28 2.27 2.25 2.24 2.21 2.21 2.14 2.13 2.09
1.44 1.22 1.28 1.61 1.36 1.37 1.21 1.30 1.18 1.31 1.20 1.28
23 18 8 6 4 31 25 22 32 17 35 15
2.22 2.57 2.97 3.11 3.45 2.00 2.17 2.45 1.91 2.58 1.86 2.62
1.38 1.08 1.17 1.53 1.27 1.22 1.22 .96 1.05 1.17 .97 1.13
30 15 10 15 4 32 29 19 31 14 28 27
1.95 2.47 2.89 2.64 3.27 1.69 2.07 2.41 1.94 2.65 2.10 2.15
1.26 .99 1.19 1.53 1.27 1.04 1.15 1.00 1.09 1.16 1.04 1.12
35 14 12 5 26 36 28 23 30 21 25 29
1.79 1.24 2.44 1.08 2.51 1.36 2.84 162 2.06 1.20 1.67 1.07 2.03 1.15 2.22 1.07 1.97 1.21 2.25 1.22 2.20 1.22 1.99 1.21
31 18 17 19 6 35 27 21 28 15 25 16
1.78 2.46 2.51 2.37 2.92 1.58 2.01 2.29 1.95 2.59 2.17 2.55
1.23 1.01 1.23 1.56 1.35 1.00 1.04 .96 1.08 1.23 1.07 1.23
30 31 32 33 34 35 36 37 38 39 40
2.05 1.94 1.71 1.69 1.69 1.66 1.58 1.57 1.56 1.52 1.50
1.27 1.38 1.17 1.11 1.04 1.12 1.09 1.11 1.00 1.04 1.10
33 7 29 34 41 24 27 38 36 39 40
1.90 3.07 2.08 1.88 1.30 2.19 2.14 1.57 1.81 1.48 1.47
1.05 1.59 1.17 .99 .72 1.12 1.16 .90 .87 .87 .85
33 3 36 26 42 17 25 38 35 39 40
1.60 3.70 1.68 2.21 1.34 2.43 2.21 1.36 1.83 1.36 1.35
.91 1.40 1.05 1.24 .81 1.15 1.14 .73 .96 .82 .81
39 27 34 31 37 20 22 38 32 41 40
1.49 2.03 1.85 1.93 1.51 2.27 2.24 1.50 1.87 1.36 1.38
38 26 34 32 42 29 30 39 33 37 41
1.41 2.03 1.72 1.76 1.24 1.90 1.80 1.31 1.76 1.42 1.24
.80 1.41 1.04 .93 .66 1.00 1.04 .71 .95 .92 .68
.88 43 1.23 .89 40 1.27 1.28 36 1.46
.68 .74 .91
–
1.03 42 1.25 1.04 43 1.23 37 1.73 –
SD
M
Spain
4.21 3.50 3.42 3.35 3.32 3.06 2.90
–
M
UK
M
41 1.50 42 1.50
R
New Zealand
1 2 3 4 5 6 7
R 39 8 40 15 38 37 13
Belgium
.61 43 1.33 .61 41 1.35 1.22 37 1.58
.77 42 1.36 .87 43 1.35 .97 33 1.87
SD
.93 1.40 1.26 1.11 .95 1.23 1.29 .95 1.09 .89 .91
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
47
was then conducted with all 12 participants regarding: the meaning of items, title, response options and questions. Respondents were also asked to give suggestions for improving the understandability of the questionnaire. As some respondents misunderstood the phrase “top of hill”, an equivalent expression was used which technically had the same meaning.
degrees of freedom (df). All analyses were carried out using SPSS 21 and AMOS 21.
2.4. Data analysis
Table 2 shows the means, standard deviations and rankings (by mean) for all ARBQ items, except the item “cross less than an hour after drinking alcohol”, which was not included in the questionnaire. The table also shows the corresponding means, SDs and rank for Belgium, New Zealand, the UK and Spain (Sullman et al., 2011, 2012; Sullman and Mann, 2009; Elliott and Baughan, 2004). In Iran, Belgium, New Zealand, England and Spain the highest mean was for the item “look both ways before crossing”. The second largest mean, in the Iranian sample was “cross at a place that is well lit when it is dark”. The second largest mean for Belgium, England, New Zealand and Spain was “check to make sure traffic has stopped before using a pedestrian crossing”, which had the third largest mean in the Iranian sample. Similar to UK and Belgium research (Sullman et al., 2012), in Iran the behaviour “play chicken by lying down in the road and waiting for cars to come along” had the lowest mean. Furthermore, the three behaviours Iranian adolescents reported least often were fairly similar to those from other countries (Elliott and Baughan,
Prior to beginning the analyses all scores for the planned protective behaviour items were reversed so that a score of 1 = “good” behaviour and 5 indicated “very bad” behaviours. This was so the items from this subscale were in the same direction as those from the other two subscales (unsafe crossing behaviour and dangerous playing on the roads). Missing data was estimated using the multiple imputation method. To explore the factor structure of the Persian version of this questionnaire, principal axis factoring (PAF) with Varimax rotations was used. Furthermore, the short version of the ARBQ was analysed using PAF and the resultant three factor model was also confirmed using CFA (confirmatory factor analysis) and a number of indices were calculated, including: the root mean square error of approximation (RMSEA), the comparative fit index (CFI), relative fit index (RFI), incremental fit index (IFI), normed fit index (NFI), Tucker–Lewis index (TLI) and chi square divided by the
3. Results 3.1. Mean comparisons between Iran and other countries
Table 3 Factor structure of the 42-item ARBQ. No.
Item (How often do you . . . )
10 25 22 21 28 23 24 30 27 29 31 26 18 4 3 9 42 12 13 5 14 15 1 17 16 32 33 34 35 36 8 40 41 11 6 2 20 19 37 38 39 7
Climb over barriers or railings that separate the road from the pavement Hold on to a moving vehicle when riding a bike Play “chicken” by deliberately running out in front of traffic Play ‘chicken’ by lying down in the road and waiting for cars to come along Run into the road to get a ball, without checking for traffic Deliberately run across the road without looking, for a dare Hold on to a moving vehicle when riding a skateboard/roller-skates/roller-blades Not notice an approaching car when playing games in the road Ride out into the road on a skateboard without thinking to check for traffic Hang around in the road talking to friends Run around in a road (e.g. when playing football or bull dog) Ride a skateboard (or roller-skates/roller-blades) on the road Run across a road without looking because you are in a hurry Cross whether traffic is coming or not, thinking the traffic should stop for you Use a mobile phone and forget to look properly Not look because you cannot hear any traffic around Walk facing the traffic when on roads without pavements Think it is OK to cross safely, but a car is coming faster than you thought Get part way across the road and then have to run the rest of the way to avoid traffic Cross from behind a stationary vehicle Have to stop quickly or turn back to avoid traffic See a small gap in traffic and “go for it” Forget to look properly because you are thinking about something else Cross between parked cars when there is a safer place to cross nearby Make traffic slow down or stop to let you cross Wear reflective clothing when crossing the road Wear reflective clothing when out on foot in the dark Wear bright or reflective clothing when riding a bike in the dark Wear a cycle helmet when riding a bike Use lights on your bike when it is dark Cross at a place that is well lit when it is dark Check to make sure traffic has stopped before using a pedestrian crossing Use a lollipop man/lady where there is one available Walking on the road rather than on the pavement Cross when you cannot see both ways very well (like on a bend or top of hill) Forget to look properly because you are talking to friends who are with you Not notice a car pulling out (say from a driveway) and walk in front of it Not bother walking to a nearby crossing to cross the road Walk in single file on roads without pavements Keep looking and listening until you get all the way across the road Look both ways before crossing Cross without waiting for the ‘green man’
Factor 1
Factor 2
Factor 3
.236 .724 .690 .689 .673 .662 .648 .630 .578 .573 .569 .550 .480 .464 .444 .421 .063 .136 .082 .035 .040 .118 .303 .224 .226 .227 .197 .218 .212 .033 .195 .249 .236 .383 .204 .280 .281 .217 .066 .291 .370 .215
.295 .034 .119 .080 .123 .182 .011 .150 .023 .282 .172 .048 .296 .333 .333 .286 .494 .456 .453 .447 .444 .433 .418 .415 .412 .097 .075 .137 .117 .020 .063 .038 .016 .378 .368 .366 .397 .366 .347 .176 .056 .303
.044 .051 .040 .093 .055 .016 .061 .048 .151 .018 .044 .108 .076 .123 .031 .157 .117 .024 .115 .033 .088 .033 .038 .148 .140 .592 .675 .665 .571 .570 .537 .524 .506 .226 .019 .034 .139 .271 .123 .335 .319 .110
48
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54 Table 4 The 21-item ARBQ factors loadings. No.
Item (How often do you . . . )
Factor loading
Factor 1. Dangerous playing on the road, alpha reliability coefficient = 0.85 25 Hold on to a moving vehicle when riding a bike 21 Play “chicken” by lying down in the road and waiting for cars to come along 22 Play “chicken” by deliberately running out in front of traffic 24 Hold on to a moving vehicle when riding a skateboard/roller-skates/roller-blades. 28 Run into the road to get a ball, without checking for traffic 23 Deliberately run across the road without looking, for a dare Ride out into the road on a skateboard without thinking to check for traffic 27 26 Ride a skateboard (or roller-skates/roller-blades) on the road
.77 .75 .74 .72 .66 .65 .62 .60
Factor 2. Planned protective behaviour, alpha reliability coefficient = 0.75 34 Wear bright or reflective clothing when riding a bike in the dark 33 Wear reflective clothing when out on foot in the dark 32 Wear reflective clothing when crossing the road 35 Wear a cycle helmet when riding a bike 36 Use lights on your bike when it is dark
.80 .79 .72 .62 .58
Factor 3. Unsafe road crossing behaviour, alpha reliability coefficient = 0.61 1 Forget to look properly because you are thinking about something else 2 Forget to look properly because you are talking to friends who are with you 12 Think it is OK to cross safely, but a car is coming faster than you thought 13 Get part way across the road and then have to run the rest of the way to avoid traffic 17 Cross between parked cars when there is a safer place to cross nearby 18 Run across a road without looking because you are in a hurry 4 Cross whether traffic is coming or not, thinking the traffic should stop for you 15 See a small gap in traffic and “go for it”
.62 .62 .54 .46 .46 .46 .40 .33
2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). The other two of the three items with the lowest mean, in the Iranian sample, were “play ‘chicken’ by deliberately running out in front of traffic “and “hold on to a moving vehicle when riding a skateboard/ roller-skates/roller-blades”, respectively. In Iranian and English adolescents the behaviour “use a lollipop man/lady where there is one available” had a comparable ranking. This item was 22nd in Iran, 26th in the UK, 4th in New Zealand, 4th in Belgium and 6th in Spain. The behaviour that the Iranian, English and Spanish students reported less often than the other two samples was “wear a cycle helmet when riding a bike” which was 31st in Iran, 27th in the UK, 26th in Spain, 3rd in New Zealand and 7th in Belgium. However, the biggest difference between Iranian adolescents and adolescents from other countries was the item “have to stop quickly or turn back to avoid traffic”, which was rated 8th in Iran, while it was the 30th, 19th, 23rd and 22nd in Belgium, the UK, New Zealand and Spain, respectively. 3.2. Factor analysis The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were used to examine the appropriateness of using exploratory factor analysis (EFA). The KMO was 0.91 and Bartlett’s test of sphericity was significant (0.30). Loadings lower than 0.3 were not reported. Factors 1, 2 and 3 explained 23.60%, 12.23% and 7.30% of the variance, respectively and the total variance explained was 43.13%. Confirmatory factor analysis (CFA) was then used on the full data set to confirm whether the 3-factor solution identified using EFA was an adequate fit for the Iranian ARBQ data. For CFA, seven fit indices evaluated the appropriateness of the model, these were: the root mean square error of approximation (RMSEA), which
Table 5 Summary of the CFA goodness-of-fit statistics for the 21 item3-factor model (n = 1111). Models a
First model fitted Second model fittedb a b
RMSEA
CFI
RFI
IFI
NFI
TLI
X 2/df
0.049 0.039
0.885 0.948
0.870 0.905
0.913 0.948
0.885 0.920
0.902 0.939
3.679 2.672
3-factor 21item ARBQ. 3-factor 21item ARBQ with 8 pairs of errors covaried.
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
should be no larger than 0.07; the comparative fit index (CFI), which should be equal or larger than 0.93; the relative fit index (RFI), which should be 0.90–1.00; the incremental fit index (IFI), which should be 0.93 to 1.00; the normed fit index (NFI), which should be 0.60–1.00; the Tucker–Lewis index (TLI), which should be equal or greater than 0.92; and chi square divided by the degrees of freedom (df), which should be less than 2.0, or at most 5.0 (Bagozzi and Yi, 2012; Oke et al., 2012; Kline, 2011). Table 4 shows the results of the CFA and that all but one of the items loads above 0.40, but all loaded above 0.30. Table 5 also shows that the initial fit of the three factor model was not ideal, but allowing 8 error pairs to co-vary resulted in a very good fit. Fig. 1 graphically presents the factor solution, which was very similar to those found by Elliott and Baughan (2004) in the UK, Sullman et al. (2011) in Spain and Sullman et al. (2012) in Belgium. Although 19 items had acceptable loadings on three factors (>0.30), the factor loadings for item 13 “get part way across the road and then have to run the rest of the way to avoid traffic” and item 15 “see a small gap in traffic and “go for it” were both lower than desired. The three factors were then tested for differences according to the demographic and descriptive variables. Table 6 shows that
49
males were more likely to be involved in dangerous playing on the road than females (p = 0 .003). Adolescents whose relatives or friends had been killed in a road crash or had experienced a road crash themselves reported playing on the road significantly more often than those who had not (p = 0.005 and p < 0.001). Dangerous playing on the road was also significantly higher among students in public schools (p = 0.008). Place of residence also influenced how likely they were to engage in dangerous playing on the road, with post-hoc tests showing that residents of large urban areas were more likely to engage in this kind of behaviour than small urban residents (p = 0.011). There were no statistically significant differences between students living in rural areas and those from other areas. In addition, post-hoc tests showed that students from schools in the small urban areas reported significantly less dangerous playing on the roads than students from schools in the northern (p < 0.001) and southern urban areas (p = 0.034), but there was no significant difference between students from schools in the north and the south of the city areas (p = 0.298). There were also no statistically significant differences on this factor by nationality, parents’ education level and grade of students.
Fig. 1. Confirmatory factor analysis of the 21-item ARBQ.
50
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
Table 7 shows that females (p = 0.008) and older students (aged 15–18 years) reported engaging less often in planned protective behaviour (p < 0.001) than males and younger students. Furthermore, adolescents who reported no previous traffic accidents reported more planned protective behaviour than those who reported prior involvement in an accident (p = 0.001). Students residing in large urban areas also reported significantly more engagement in planned protective behaviour than those from small urban areas (p = 0.001), but there were no significant differences between rural students and those from small and large urban areas. Engagement in planned protective behaviour was significantly lower amongst respondents from schools in small urban areas, compared with those from schools in the northern (p = 0.003) and southern (p = 0.008) urban areas. However, there was no significant difference between participants from schools in the northern and southern urban areas (p = 0.887). Finally, there were no statistically significant differences found between planned protective behaviour and the other variables. Table 8 presents the relationships the unsafe crossing behaviour factor had with the demographic and descriptive variables. Students in the 15–18 years old age range reported significantly more often engaging in unsafe crossing behaviour than the younger age group (p = 0.031). Respondents whose friends and relatives had lost their lives in a road crash also had Table 6 Subgroup comparisons between scores gained in factor 1 analysis (dangerous playing in the road) through t-test and ANOVA. Variable Gender Males Females Age 13–14 15–18 Nationality Iranian Afghan Living area Large urban area Small urban area Rural area Mother's education level Illiterate Elementary school Guidance school High school Diploma University Father's education level Illiterate Elementary school Guidance school High school Diploma University Relatives or friends killed in a road crash Yes No Traffic accident history Yes No Family motorized vehicle Yes No Kind of school Public Private Location of school The north of Tehran (large urban) The south of Tehran (large urban) Small urban *
Significant at the 0.05 level.
Mean
SD
1.57 1.44
1.70 1.56
1.60 1.55
0.73 0.77
1.57 1.51
0.76 0.63
1.60 1.43 1.65
0.79 0.61 0.74
1.69 1.53 1.51 1.72 1.58 1.55
.84 .67 .67 .85 .79 .76
1.54 1.46 1.49 1.74 1.62 1.54
0.71 0.68 0.62 0.83 0.82 0.75
p-value 0.003*
0.259
0.535
0.011*
0.370
0.064
*
0.005 1.63 1.51
0.77 0.74
1.73 1.48
0.82 0.71
1.58 1.52
0.77 0.71
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
The road user behaviour of school students in Iran Amir Reza Nabipour a , Nouzar Nakhaee b , Narges Khanjani a, *, Hossein Zirak Moradlou c , Mark J.M. Sullman d a
Department of Biostatistics and Epidemiology, Faculty of Public Health, Kerman University of Medical Sciences, Haft Bagh Alavi Blvd., Kerman, Iran Neuroscience Research Center, Kerman University of Medical Sciences, Kerman, Iran c Consultation Center, Tehran Education and Training Organization, Tehran, Iran d Driving Research Group, Cranfield University, UK b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 29 March 2014 Received in revised form 28 October 2014 Accepted 10 November 2014 Available online xxx
The present study developed a Persian version of the Adolescent Road User Behaviour Questionnaire (ARBQ) and investigated the psychometric properties of the scale in a sample of school students in the province of Tehran (Iran). In total 1111 adolescents completed the Persian version of the ARBQ. Exploratory factor analysis, using the shortened 21-item version of the scale revealed the presence of three reliable factors which were also supported using confirmatory factor analysis. According to this research, engagement in dangerous playing in the road was significantly higher among males, residents of large urban areas, students from private schools, students in the south of Tehran, those who reported relatives or friends had been killed in a road crash and those with a personal history of road accidents. Moreover, older adolescents, those who reported relatives or friends having been killed in a road crash and those with a traffic accident history reported higher involvement in unsafe crossing behaviour. Females, older adolescents, residents of small urban areas, students from schools in small urban areas and those with an accident history also reported less frequent engagement in planned protective behaviours. This study confirms that the ARBQ is a useful framework for investigating adolescents’ on-road behaviours in Iran. This research also showed that adolescents put themselves at risk by engaging in hazardous behaviours. As is the case in most countries, this study revealed the need for interventions, such as education and enforcement to improve the on-road safety culture amongst Iranian adolescents. ã 2014 Elsevier Ltd. All rights reserved.
Keywords: Psychometric assessment Adolescents behaviour Pedestrians ARBQ Road safety
1. Introduction Road traffic accidents pose a major public health problem by causing a large number of injuries, disabilities and fatalities, especially in the low and medium-income countries (Peden et al., 2004; Olukoga et al., 2011; Olukoga, 2003). This is also the case for Iran, which in 2010 recorded a total of 414,161 injuries and 23,249 deaths. The number of fatalities translates into a rate of 12.4/10,000 registered vehicles, which is substantially higher than in most developed countries, such as New Zealand (1.2/10,000 registered vehicles), Austria (1.0/10,000 registered vehicles), the USA (1.3/10,000 registered vehicles) and is more than twenty times higher than the 0.6/10,000 registered vehicles reported by the UK (NZTA, 2011; Bahadorimonfared et al., 2013). Furthermore, the Iranian Forensic Medicine Organization (2012)
* Corresponding author. Tel.: +98 34 3132 5102. E-mail address: [email protected] (N. Khanjani). http://dx.doi.org/10.1016/j.aap.2014.11.009 0001-4575/ ã 2014 Elsevier Ltd. All rights reserved.
reported that from 2006–2011 a total of 120,070 people died on Iranian roads and of these 7565 (6.3%) were adolescents aged between 11 and 17 years old. A number of studies have reported that adolescents have an inflated risk of being killed or injured on the roads (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). One of the reasons for this inflated risk is that adolescents carry out a variety of unsafe and potentially risky on-road behaviours which increases the risk of being killed or injured on the road (West et al., 1998; Poudel-Tandukar et al., 2006; Campbell and Keegan, 2000; Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). One important step for improving the safety of Iranian adolescents is to understand the behaviours that put them at greater risk of being killed or injured on the road. Although there is currently no widely agreed upon framework for investigating the on-road behaviours of adolescents, one which has been utilised in several countries across the world is the Adolescent Road User Behaviour Questionnaire (ARBQ; Elliott and Baughan, 2004). Using
44
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
data from adolescent school students in the UK, Elliott and Baughan investigated the psychometric properties of their 43-item ARBQ and found three factors best described their data. The first factor consisted of items relating to crossing the road in an unsafe manner (e.g. run across a road without looking because you are in a hurry), which they called “unsafe crossing behaviour”. The second factor, “dangerous playing in the road” consisted of items involving playing on the road (e.g. hold on to a moving vehicle when riding a bike). In contrast the third factor “planned protective behaviours” was comprised of safety behaviours that reduce the chance of being killed or injured on the road (e.g. wear bright or reflective clothing when out on foot in the dark). The items of the ARBQ measure not only different types of pedestrian behaviours but also a small number involving: bicycles, skateboards and rollerblades (e.g. use lights on your bike when it is dark). Elliott and Baughan also developed a short 21-item version of the scale, which reflected the same three factors and was developed for use in conjunction with other psychometric scales. The psychometric properties of the ARBQ have also been investigated in New Zealand (Sullman and Mann, 2009), Spain (Sullman et al., 2011), Belgium (Sullman et al., 2012) and France (Abou et al., 2008). These studies all supported the presence of the three factors and the three factors were also supported for the 21item version of the scale in Spain and Belgium using confirmatory factor analysis. However, these findings contrast with the New Zealand research which produced a 19-item version of the scale using exploratory factor analysis (Sullman and Mann, 2009). Although there were considerable differences between the four studies which have used the ARBQ, in terms of different: countries, populations, cultures, ethnic groups, languages, traffic conditions and rules, road traffic environments and population densities, there has been a surprising number of consistent findings. For instance, in terms of gender, in all four studies male students reported more frequent engagement in playing on the road than female adolescents (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). Although, the New Zealand study did not show statistically significant differences between gender and unsafe crossing behaviour (Sullman and Mann, 2009), the other three studies all found that males were more likely to be involved in these types of potentially risky behaviours (Sullman et al., 2011, 2012). However, none of the previous studies found a significant relationship between planned protective behaviour and gender. There were also similarities in the relationships the different ARBQ factors had with age. For example, the English and Spanish studies found that older adolescents reported more unsafe crossing behaviour, while also engaging less often in planned protective behaviour and dangerous playing in the road (Elliott and Baughan, 2004; Sullman et al., 2011). In Belgium, adolescents who were 11–12 years of age reported a lower frequency of unsafe road crossing behaviour, than those aged 13–14 years and greater than 15 years of age (Sullman et al., 2012). Finally, the Belgian research also found that younger respondents reported more planned protective behaviour than older adolescents (Sullman et al., 2012), as was also the case in the UK and Spanish studies (Sullman et al., 2011, 2012). However, as the New Zealand research used very different age categories it was not possible to compare their results with the other ARBQ studies (Sullman and Mann, 2009). As well as age and gender, the ARBQ factors have been found to be related to an adolescent’s place of residence. Elliott and Baughan (2004) found that students from large urban areas reported engaging more often in unsafe road crossing behaviours than those from rural and small urban areas. Furthermore, adolescents living in large urban areas reported less frequently engaging in dangerous playing on the roads than adolescents from rural areas, while rural residents reported more planned protective behaviours than those
from small and large urban areas. However, in contrast to the UK research there were no statistically significant differences between living areas and the ARBQ factors in New Zealand, Belgium or Spain (Sullman and Mann, 2009; Sullman et al., 2011). As would be expected, given the fact that the ARBQ investigates behaviours thought to be important for road safety, previous research has found that adolescents who self-reported previous involvement in a road crash also reported more frequently engaging in the two types of potentially risky behaviours and also, unexpectedly, planned protective behaviours (Sullman et al., 2012). Also surprising was the fact that thus far only one study has investigated the relationship the ARBQ factors have with prior crash involvement. As there is only one study, this relationship may be an anomaly or confined solely to Belgium. Therefore, it is important that research be conducted to test this relationship in a novel sample and preferably in a novel country and culture. Another important gap in the literature is that all previous research using the ARBQ has been conducted in developed countries. This is a particularly important limitation as the size of the road safety problem, in terms of injuries and fatalities, is considerably larger in less developed countries. Therefore, the present study investigated the psychometric properties of the ARBQ in Iran, a developing country with a non-European culture and background. The replication of these findings is a critical step to confirm the suitability of the ARBQ as a framework to measure road user behaviours in this at-risk age group. More specifically, the present study investigated the factor structure of the ARBQ in a large sample of Iranian school students using both exploratory and confirmatory factor analysis for the long and short versions of the ARBQ. The study also investigated the relationships the resultant factors had with age, gender, place of residence, along with the individuals' accident history and those of their friends and family. 2. Methods 2.1. Materials The ARBQ (Adolescent Road-User Behaviour Questionnaire) is a self-report tool for investigating how often participants engage in 43 different road-user behaviours. For example, “How often do you not look because you can't hear any traffic coming?” Responses are made on a five point Likert Scale (1 = never to 5 = very often). Previous studies have found that the questionnaire is measuring three latent variables: “unsafe road crossing behaviour”, “dangerous playing on the road” and “planned protective behaviour” (Elliott and Baughan, 2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). These three factors had internal reliability scores of .89, .85 and .76, respectively. The questionnaire also asked about the participants’ demographic and descriptive details, including: age, gender, nationality, living area, mother’s education level, father’s education level, school type (public or private), school grade, location of the school, along with whether they (or their friends and/or family) had been involved in a road crash. 2.2. Setting and participants The present cross sectional study was carried out in secondary schools in the province of Tehran, Iran. The sample was comprised of 1200 students from large and small urban areas which were randomly selected. Tehran city was selected as the large urban area and Pishva was selected as the small urban area within the same province. In the 2011 Nationwide Iranian Census Tehran city had a population of 8,244,535 while Pishva had a population of 75,454 (Iranian Statistics Center, 2011).
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
From Tehran’s 22 municipal districts, two municipal districts were randomly selected, one from the north (high socioeconomic area) and the other from the south (low socioeconomic area). Four public and four private schools (8 schools) were randomly selected from each district. Half of these schools were for boys’ and the other half were girls’ schools. In the town of Pishva four public secondary schools (two boys’ and two girls’ schools) were randomly selected. In total, 20 schools were included in the study, half of which were junior high schools (7th, 8th, 9th grade) and the other half were high schools (10th, 11th, 12th grade). All random
45
selection took place the same way for all aspects of the study. This was to allocate each province, school and participant a number and then to use SPSS to generate the required quantity of random numbers. Once the schools had been selected, permission was sought from the authorities at each school before starting the study. From each school, 120 students from grade 7 to 12 and between 13 and 18 years old were randomly chosen. All students consented to participate in the survey. Participants were guaranteed anonymity and confidentiality of their answers. All students
Table 1 The characteristics of the students participating in this study. Variable
Frequency
percentage
Gender Males Females
557 554
50.1 49.9
Age 13–14 15–18
378 733
34 66
Grade 7 8 9 10 11 12
173 204 188 193 185 168
15.6 18.4 16.9 17.4 16.7 15.1
1043 68
93.9 6.1
Living area Large urban area Small urban area Rural area
840 215 56
75.6 19.4 5
Mother’s education level Illiterate Elementary school (grade 1–5) Guidance school (grade 6–8) High school (grade 9–12) Diploma Academic level
51 143 139 63 384 331
4.6 12.9 12.5 5.7 34.6 29.8
Father’s education level Illiterate Elementary school Guidance school High school Diploma Academic level
45 115 138 85 353 375
4.1 10.4 12.4 7.7 31.8 33.8
Relatives or friends killed ina road crash Yes No
541 570
48.7 51.3
Traffic accident history Yes No
385 726
34.7 65.3
Family motorized vehicle Yes No
882 229
79.4 20.6
Kind of school Public Private
679 432
61.1 38.9
Location of school (according to socioeconomic status) The north of Tehran (large urban, high socioeconomic status) The south of Tehran (large urban, low socioeconomic status) Small urban
412 458 241
37.1 41.2 21.7
Nationality Iranian Afghan
46
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
2.3. Cross cultural adaptation
took 25–30 min to answer the Persian version of ARBQ. Although all 1200 randomly selected students agreed to participate in this project, 89 students did not complete the survey, leaving a total of 1111 participants (response rate of 92.6%). Of the 1111 students who answered the questionnaire, 557 (50.1%) were male and 554 (49.9%) female, with an average age of 15.45 years old (SD = 1.69, range 13–18 years old). The nationality of these students was mainly Iranian and most lived in Tehran city (Table 1). The adolescence period is generally divided into three periods, including: early (aged 11–14), middle (aged 15–18) and late (aged 18–21) periods (Lerner and Laurence, 1999). In the current study, the number of participants in ages 13–14 and ages 15–18 were 378 (34%) and 733 (66%), respectively. Participants were also asked whether they had been involved in an accident as a pedestrian or while riding a bicycle. Over a third (34.7%, 385) reported that they had been involved in an accident on the road. The characteristics of the participants can be seen in Table 1.
2.3.1. Translation First, forward translations were carried out independently by two native translators. These translations were compared and discussed by researchers and the draft Persian version was prepared. Then, two bilingual native English translators independently translated the Persian version back into English. The translators were not aware of the content of the initial English version. Comparisons were then made between the backward translations and the original version, but no major differences were identified. The draft Persian-language version of the ARBQ was developed by consensus between the translators and researchers. At this point, as drinking alcohol is religiously prohibited for Muslims there searchers made the decision to exclude the item “cross less than an hour after drinking alcohol”. 2.3.2. Cognitive testing The draft Persian version of the questionnaire was then tested on 12 students with an age range of 13–18 years old. An interview
Table 2 ARBQ items means and standard deviation of Iran, Belgium, New Zealand, UK and Spain. No. Item (How often do you. .)
Iran
14 5 17 12 19 16 7 2 20 6 34 1 42 36 41 33 3 11 10 9 18 4 32 35 26 30 23 29 31 27 28 25 24 22 21 –
Look both ways before crossing Cross at a place that is well lit when it is dark Check to make sure traffic has stopped before using a pedestrian crossing See a small gap in traffic and “go for it” Keep looking and listening until you get all the way across the road Walk in single file on roads without pavements Get part way across the road and then have to run the rest of the way to avoid traffic Have to stop quickly or turn back to avoid traffic Cross from behind a stationary vehicle Cross between parked cars when there is a safer place to cross nearby Think it is OK to cross safely, but a car is coming faster than you thought Not bother walking to a nearby crossing to cross the road Make traffic slow down or stop to let you cross Cross without waiting for the ‘green man’ Forget to look properly because you are talking to friends who are with you Not notice a car pulling out (say from a driveway) and walk in front of it Cross when you cannot see both ways very well (like on a bend or top of hill) Wear bright or reflective clothing when riding a bike in the dark Forget to look properly because you are thinking about something else Walk facing the traffic when on roads without pavements Use lights on your bike when it is dark Use a lollipop man/lady where there is one available Wear reflective clothing when out on foot in the dark Use a mobile phone and forget to look properly Walking on the road rather than on the pavement Climb over barriers or railings that separate the road from the pavement Not look because you cannot hear any traffic around Run across a road without looking because you are in a hurry Cross whether traffic is coming or not, thinking the traffic should stop for you Wear reflective clothing when crossing the road Wear a cycle helmet when riding a bike Ride on a skateboard (or roller-skates/roller-blades) on the road Not notice an approaching car when playing games in the road Deliberately run across the road without looking, for a dare Hang around in the road talking to friends Run around in a road (e.g. when playing foot ballor bull dog) Ride out into the road on a skateboard without thinking to check for traffic Run into the road to get a ball, without checking for traffic. Hold on to a moving vehicle when riding a bike Hold on to a moving vehicle when riding a skateboard/roller-skates/rollerblades. Play “chicken” by deliberately running out in front of traffic Play ‘chicken’ by lying down in the road and waiting for cars to come along Cross less than an hour after drinking alcohol
R
SD
SD
R
M
R
M
SD
1.13 1 4.17 .95 1 4.17 1.35 5 3.36 1.06 7 3.13 1.34 2 3.87 1.05 2 3.73 1.36 20 2.56 1.13 6 3.14 1.4 3 3.55 1.14 5 3.24 1.41 10 2.86 1.29 20 2.40 1.38 11 2.74 1.04 9 3.01
.95 1.08 1.15 1.08 1.17 1.16 1.04
1 3 2 8 4 11 6
4.08 3.32 3.46 2.69 3.26 2.54 2.82
1.07 1.16 1.30 1.30 1.26 1.38 1.14
1 3 2 13 4 10 8
4.07 3.48 3.76 2.73 3.34 2.81 2.88
1.08 1.18 1.15 1.13 1.24 1.41 1.04
8 9 10 11 12 13 14 15 16 17
2.87 2.77 2.68 2.59 2.58 2.57 2.46 2.45 2.43 2.38
1.31 30 2.06 .91 1.27 14 2.64 1.00 1.30 12 2.70 .99 1.17 26 2.16 .94 1.36 13 2.66 1.05 1.36 9 2.85 1.19 1.4 21 2.50 1.23 1.35 19 2.57 1.16 1.11 28 2.13 .90 1.29 16 2.59 1.01
23 12 11 21 8 24 18 13 22 16
2.23 2.83 2.88 2.38 3.10 2.22 2.42 2.80 2.30 2.45
.94 19 2.34 1.05 16 2.35 1.03 9 2.66 .92 18 2.34 1.13 7 2.72 1.05 15 2.43 1.23 17 2.35 1.06 10 2.65 1.03 24 2.20 .94 13 2.49
1.07 1.17 1.16 1.12 1.22 1.24 1.17 1.15 1.03 1.10
22 11 7 24 9 14 12 5 20 23
2.27 2.79 2.92 2.17 2.85 2.66 2.79 3.05 2.36 2.25
1.01 1.12 1.07 .98 1.14 1.14 1.12 1.12 1.09 .96
18 19 20 21 22 23 24 25 26 27 28 29
2.35 2.33 2.32 2.28 2.27 2.25 2.24 2.21 2.21 2.14 2.13 2.09
1.44 1.22 1.28 1.61 1.36 1.37 1.21 1.30 1.18 1.31 1.20 1.28
23 18 8 6 4 31 25 22 32 17 35 15
2.22 2.57 2.97 3.11 3.45 2.00 2.17 2.45 1.91 2.58 1.86 2.62
1.38 1.08 1.17 1.53 1.27 1.22 1.22 .96 1.05 1.17 .97 1.13
30 15 10 15 4 32 29 19 31 14 28 27
1.95 2.47 2.89 2.64 3.27 1.69 2.07 2.41 1.94 2.65 2.10 2.15
1.26 .99 1.19 1.53 1.27 1.04 1.15 1.00 1.09 1.16 1.04 1.12
35 14 12 5 26 36 28 23 30 21 25 29
1.79 1.24 2.44 1.08 2.51 1.36 2.84 162 2.06 1.20 1.67 1.07 2.03 1.15 2.22 1.07 1.97 1.21 2.25 1.22 2.20 1.22 1.99 1.21
31 18 17 19 6 35 27 21 28 15 25 16
1.78 2.46 2.51 2.37 2.92 1.58 2.01 2.29 1.95 2.59 2.17 2.55
1.23 1.01 1.23 1.56 1.35 1.00 1.04 .96 1.08 1.23 1.07 1.23
30 31 32 33 34 35 36 37 38 39 40
2.05 1.94 1.71 1.69 1.69 1.66 1.58 1.57 1.56 1.52 1.50
1.27 1.38 1.17 1.11 1.04 1.12 1.09 1.11 1.00 1.04 1.10
33 7 29 34 41 24 27 38 36 39 40
1.90 3.07 2.08 1.88 1.30 2.19 2.14 1.57 1.81 1.48 1.47
1.05 1.59 1.17 .99 .72 1.12 1.16 .90 .87 .87 .85
33 3 36 26 42 17 25 38 35 39 40
1.60 3.70 1.68 2.21 1.34 2.43 2.21 1.36 1.83 1.36 1.35
.91 1.40 1.05 1.24 .81 1.15 1.14 .73 .96 .82 .81
39 27 34 31 37 20 22 38 32 41 40
1.49 2.03 1.85 1.93 1.51 2.27 2.24 1.50 1.87 1.36 1.38
38 26 34 32 42 29 30 39 33 37 41
1.41 2.03 1.72 1.76 1.24 1.90 1.80 1.31 1.76 1.42 1.24
.80 1.41 1.04 .93 .66 1.00 1.04 .71 .95 .92 .68
.88 43 1.23 .89 40 1.27 1.28 36 1.46
.68 .74 .91
–
1.03 42 1.25 1.04 43 1.23 37 1.73 –
SD
M
Spain
4.21 3.50 3.42 3.35 3.32 3.06 2.90
–
M
UK
M
41 1.50 42 1.50
R
New Zealand
1 2 3 4 5 6 7
R 39 8 40 15 38 37 13
Belgium
.61 43 1.33 .61 41 1.35 1.22 37 1.58
.77 42 1.36 .87 43 1.35 .97 33 1.87
SD
.93 1.40 1.26 1.11 .95 1.23 1.29 .95 1.09 .89 .91
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
47
was then conducted with all 12 participants regarding: the meaning of items, title, response options and questions. Respondents were also asked to give suggestions for improving the understandability of the questionnaire. As some respondents misunderstood the phrase “top of hill”, an equivalent expression was used which technically had the same meaning.
degrees of freedom (df). All analyses were carried out using SPSS 21 and AMOS 21.
2.4. Data analysis
Table 2 shows the means, standard deviations and rankings (by mean) for all ARBQ items, except the item “cross less than an hour after drinking alcohol”, which was not included in the questionnaire. The table also shows the corresponding means, SDs and rank for Belgium, New Zealand, the UK and Spain (Sullman et al., 2011, 2012; Sullman and Mann, 2009; Elliott and Baughan, 2004). In Iran, Belgium, New Zealand, England and Spain the highest mean was for the item “look both ways before crossing”. The second largest mean, in the Iranian sample was “cross at a place that is well lit when it is dark”. The second largest mean for Belgium, England, New Zealand and Spain was “check to make sure traffic has stopped before using a pedestrian crossing”, which had the third largest mean in the Iranian sample. Similar to UK and Belgium research (Sullman et al., 2012), in Iran the behaviour “play chicken by lying down in the road and waiting for cars to come along” had the lowest mean. Furthermore, the three behaviours Iranian adolescents reported least often were fairly similar to those from other countries (Elliott and Baughan,
Prior to beginning the analyses all scores for the planned protective behaviour items were reversed so that a score of 1 = “good” behaviour and 5 indicated “very bad” behaviours. This was so the items from this subscale were in the same direction as those from the other two subscales (unsafe crossing behaviour and dangerous playing on the roads). Missing data was estimated using the multiple imputation method. To explore the factor structure of the Persian version of this questionnaire, principal axis factoring (PAF) with Varimax rotations was used. Furthermore, the short version of the ARBQ was analysed using PAF and the resultant three factor model was also confirmed using CFA (confirmatory factor analysis) and a number of indices were calculated, including: the root mean square error of approximation (RMSEA), the comparative fit index (CFI), relative fit index (RFI), incremental fit index (IFI), normed fit index (NFI), Tucker–Lewis index (TLI) and chi square divided by the
3. Results 3.1. Mean comparisons between Iran and other countries
Table 3 Factor structure of the 42-item ARBQ. No.
Item (How often do you . . . )
10 25 22 21 28 23 24 30 27 29 31 26 18 4 3 9 42 12 13 5 14 15 1 17 16 32 33 34 35 36 8 40 41 11 6 2 20 19 37 38 39 7
Climb over barriers or railings that separate the road from the pavement Hold on to a moving vehicle when riding a bike Play “chicken” by deliberately running out in front of traffic Play ‘chicken’ by lying down in the road and waiting for cars to come along Run into the road to get a ball, without checking for traffic Deliberately run across the road without looking, for a dare Hold on to a moving vehicle when riding a skateboard/roller-skates/roller-blades Not notice an approaching car when playing games in the road Ride out into the road on a skateboard without thinking to check for traffic Hang around in the road talking to friends Run around in a road (e.g. when playing football or bull dog) Ride a skateboard (or roller-skates/roller-blades) on the road Run across a road without looking because you are in a hurry Cross whether traffic is coming or not, thinking the traffic should stop for you Use a mobile phone and forget to look properly Not look because you cannot hear any traffic around Walk facing the traffic when on roads without pavements Think it is OK to cross safely, but a car is coming faster than you thought Get part way across the road and then have to run the rest of the way to avoid traffic Cross from behind a stationary vehicle Have to stop quickly or turn back to avoid traffic See a small gap in traffic and “go for it” Forget to look properly because you are thinking about something else Cross between parked cars when there is a safer place to cross nearby Make traffic slow down or stop to let you cross Wear reflective clothing when crossing the road Wear reflective clothing when out on foot in the dark Wear bright or reflective clothing when riding a bike in the dark Wear a cycle helmet when riding a bike Use lights on your bike when it is dark Cross at a place that is well lit when it is dark Check to make sure traffic has stopped before using a pedestrian crossing Use a lollipop man/lady where there is one available Walking on the road rather than on the pavement Cross when you cannot see both ways very well (like on a bend or top of hill) Forget to look properly because you are talking to friends who are with you Not notice a car pulling out (say from a driveway) and walk in front of it Not bother walking to a nearby crossing to cross the road Walk in single file on roads without pavements Keep looking and listening until you get all the way across the road Look both ways before crossing Cross without waiting for the ‘green man’
Factor 1
Factor 2
Factor 3
.236 .724 .690 .689 .673 .662 .648 .630 .578 .573 .569 .550 .480 .464 .444 .421 .063 .136 .082 .035 .040 .118 .303 .224 .226 .227 .197 .218 .212 .033 .195 .249 .236 .383 .204 .280 .281 .217 .066 .291 .370 .215
.295 .034 .119 .080 .123 .182 .011 .150 .023 .282 .172 .048 .296 .333 .333 .286 .494 .456 .453 .447 .444 .433 .418 .415 .412 .097 .075 .137 .117 .020 .063 .038 .016 .378 .368 .366 .397 .366 .347 .176 .056 .303
.044 .051 .040 .093 .055 .016 .061 .048 .151 .018 .044 .108 .076 .123 .031 .157 .117 .024 .115 .033 .088 .033 .038 .148 .140 .592 .675 .665 .571 .570 .537 .524 .506 .226 .019 .034 .139 .271 .123 .335 .319 .110
48
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54 Table 4 The 21-item ARBQ factors loadings. No.
Item (How often do you . . . )
Factor loading
Factor 1. Dangerous playing on the road, alpha reliability coefficient = 0.85 25 Hold on to a moving vehicle when riding a bike 21 Play “chicken” by lying down in the road and waiting for cars to come along 22 Play “chicken” by deliberately running out in front of traffic 24 Hold on to a moving vehicle when riding a skateboard/roller-skates/roller-blades. 28 Run into the road to get a ball, without checking for traffic 23 Deliberately run across the road without looking, for a dare Ride out into the road on a skateboard without thinking to check for traffic 27 26 Ride a skateboard (or roller-skates/roller-blades) on the road
.77 .75 .74 .72 .66 .65 .62 .60
Factor 2. Planned protective behaviour, alpha reliability coefficient = 0.75 34 Wear bright or reflective clothing when riding a bike in the dark 33 Wear reflective clothing when out on foot in the dark 32 Wear reflective clothing when crossing the road 35 Wear a cycle helmet when riding a bike 36 Use lights on your bike when it is dark
.80 .79 .72 .62 .58
Factor 3. Unsafe road crossing behaviour, alpha reliability coefficient = 0.61 1 Forget to look properly because you are thinking about something else 2 Forget to look properly because you are talking to friends who are with you 12 Think it is OK to cross safely, but a car is coming faster than you thought 13 Get part way across the road and then have to run the rest of the way to avoid traffic 17 Cross between parked cars when there is a safer place to cross nearby 18 Run across a road without looking because you are in a hurry 4 Cross whether traffic is coming or not, thinking the traffic should stop for you 15 See a small gap in traffic and “go for it”
.62 .62 .54 .46 .46 .46 .40 .33
2004; Sullman and Mann, 2009; Sullman et al., 2011, 2012). The other two of the three items with the lowest mean, in the Iranian sample, were “play ‘chicken’ by deliberately running out in front of traffic “and “hold on to a moving vehicle when riding a skateboard/ roller-skates/roller-blades”, respectively. In Iranian and English adolescents the behaviour “use a lollipop man/lady where there is one available” had a comparable ranking. This item was 22nd in Iran, 26th in the UK, 4th in New Zealand, 4th in Belgium and 6th in Spain. The behaviour that the Iranian, English and Spanish students reported less often than the other two samples was “wear a cycle helmet when riding a bike” which was 31st in Iran, 27th in the UK, 26th in Spain, 3rd in New Zealand and 7th in Belgium. However, the biggest difference between Iranian adolescents and adolescents from other countries was the item “have to stop quickly or turn back to avoid traffic”, which was rated 8th in Iran, while it was the 30th, 19th, 23rd and 22nd in Belgium, the UK, New Zealand and Spain, respectively. 3.2. Factor analysis The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were used to examine the appropriateness of using exploratory factor analysis (EFA). The KMO was 0.91 and Bartlett’s test of sphericity was significant (0.30). Loadings lower than 0.3 were not reported. Factors 1, 2 and 3 explained 23.60%, 12.23% and 7.30% of the variance, respectively and the total variance explained was 43.13%. Confirmatory factor analysis (CFA) was then used on the full data set to confirm whether the 3-factor solution identified using EFA was an adequate fit for the Iranian ARBQ data. For CFA, seven fit indices evaluated the appropriateness of the model, these were: the root mean square error of approximation (RMSEA), which
Table 5 Summary of the CFA goodness-of-fit statistics for the 21 item3-factor model (n = 1111). Models a
First model fitted Second model fittedb a b
RMSEA
CFI
RFI
IFI
NFI
TLI
X 2/df
0.049 0.039
0.885 0.948
0.870 0.905
0.913 0.948
0.885 0.920
0.902 0.939
3.679 2.672
3-factor 21item ARBQ. 3-factor 21item ARBQ with 8 pairs of errors covaried.
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
should be no larger than 0.07; the comparative fit index (CFI), which should be equal or larger than 0.93; the relative fit index (RFI), which should be 0.90–1.00; the incremental fit index (IFI), which should be 0.93 to 1.00; the normed fit index (NFI), which should be 0.60–1.00; the Tucker–Lewis index (TLI), which should be equal or greater than 0.92; and chi square divided by the degrees of freedom (df), which should be less than 2.0, or at most 5.0 (Bagozzi and Yi, 2012; Oke et al., 2012; Kline, 2011). Table 4 shows the results of the CFA and that all but one of the items loads above 0.40, but all loaded above 0.30. Table 5 also shows that the initial fit of the three factor model was not ideal, but allowing 8 error pairs to co-vary resulted in a very good fit. Fig. 1 graphically presents the factor solution, which was very similar to those found by Elliott and Baughan (2004) in the UK, Sullman et al. (2011) in Spain and Sullman et al. (2012) in Belgium. Although 19 items had acceptable loadings on three factors (>0.30), the factor loadings for item 13 “get part way across the road and then have to run the rest of the way to avoid traffic” and item 15 “see a small gap in traffic and “go for it” were both lower than desired. The three factors were then tested for differences according to the demographic and descriptive variables. Table 6 shows that
49
males were more likely to be involved in dangerous playing on the road than females (p = 0 .003). Adolescents whose relatives or friends had been killed in a road crash or had experienced a road crash themselves reported playing on the road significantly more often than those who had not (p = 0.005 and p < 0.001). Dangerous playing on the road was also significantly higher among students in public schools (p = 0.008). Place of residence also influenced how likely they were to engage in dangerous playing on the road, with post-hoc tests showing that residents of large urban areas were more likely to engage in this kind of behaviour than small urban residents (p = 0.011). There were no statistically significant differences between students living in rural areas and those from other areas. In addition, post-hoc tests showed that students from schools in the small urban areas reported significantly less dangerous playing on the roads than students from schools in the northern (p < 0.001) and southern urban areas (p = 0.034), but there was no significant difference between students from schools in the north and the south of the city areas (p = 0.298). There were also no statistically significant differences on this factor by nationality, parents’ education level and grade of students.
Fig. 1. Confirmatory factor analysis of the 21-item ARBQ.
50
A.R. Nabipour et al. / Accident Analysis and Prevention 75 (2015) 43–54
Table 7 shows that females (p = 0.008) and older students (aged 15–18 years) reported engaging less often in planned protective behaviour (p < 0.001) than males and younger students. Furthermore, adolescents who reported no previous traffic accidents reported more planned protective behaviour than those who reported prior involvement in an accident (p = 0.001). Students residing in large urban areas also reported significantly more engagement in planned protective behaviour than those from small urban areas (p = 0.001), but there were no significant differences between rural students and those from small and large urban areas. Engagement in planned protective behaviour was significantly lower amongst respondents from schools in small urban areas, compared with those from schools in the northern (p = 0.003) and southern (p = 0.008) urban areas. However, there was no significant difference between participants from schools in the northern and southern urban areas (p = 0.887). Finally, there were no statistically significant differences found between planned protective behaviour and the other variables. Table 8 presents the relationships the unsafe crossing behaviour factor had with the demographic and descriptive variables. Students in the 15–18 years old age range reported significantly more often engaging in unsafe crossing behaviour than the younger age group (p = 0.031). Respondents whose friends and relatives had lost their lives in a road crash also had Table 6 Subgroup comparisons between scores gained in factor 1 analysis (dangerous playing in the road) through t-test and ANOVA. Variable Gender Males Females Age 13–14 15–18 Nationality Iranian Afghan Living area Large urban area Small urban area Rural area Mother's education level Illiterate Elementary school Guidance school High school Diploma University Father's education level Illiterate Elementary school Guidance school High school Diploma University Relatives or friends killed in a road crash Yes No Traffic accident history Yes No Family motorized vehicle Yes No Kind of school Public Private Location of school The north of Tehran (large urban) The south of Tehran (large urban) Small urban *
Significant at the 0.05 level.
Mean
SD
1.57 1.44
1.70 1.56
1.60 1.55
0.73 0.77
1.57 1.51
0.76 0.63
1.60 1.43 1.65
0.79 0.61 0.74
1.69 1.53 1.51 1.72 1.58 1.55
.84 .67 .67 .85 .79 .76
1.54 1.46 1.49 1.74 1.62 1.54
0.71 0.68 0.62 0.83 0.82 0.75
p-value 0.003*
0.259
0.535
0.011*
0.370
0.064
*
0.005 1.63 1.51
0.77 0.74
1.73 1.48
0.82 0.71
1.58 1.52
0.77 0.71
