Accident Analysis and Prevention 79 (2015) 145–151
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Crash risk and aberrant driving behaviors among bus drivers: The role of personality and attitudes towards traffic safety Luca Mallia a,b, * , Lambros Lazuras a , Cristiano Violani c , Fabio Lucidi a a b c
Department of Psychology of Development and Socialization Processes, Sapienza University of Rome, Via Dei Marsi 78, 00185 Rome, Italy Department of Movement, Human and Health Sciences, University of Rome “Foro Italico”, P.za Lauro de Bosis 15, 00135 Rome, Italy Department of Psychology, Sapienza University of Rome, Via Dei Marsi 78, 00185, Rome, Italy
A R T I C L E I N F O
A B S T R A H A C T
Article history: Received 3 December 2014 Received in revised form 25 February 2015 Accepted 23 March 2015 Available online xxx
Several studies have shown that personality traits and attitudes toward traffic safety predict aberrant driving behaviors and crash involvement. However, this process has not been adequately investigated in professional drivers, such as bus drivers. The present study used a personality–attitudes model to assess whether personality traits predicted aberrant self-reported driving behaviors (driving violations, lapses, and errors) both directly and indirectly, through the effects of attitudes towards traffic safety in a large sample of bus drivers. Additionally, the relationship between aberrant self-reported driving behaviors and crash risk was also assessed. Three hundred and one bus drivers (mean age = 39.1, SD = 10.7 years) completed a structured and anonymous questionnaire measuring personality traits, attitudes toward traffic safety, self-reported aberrant driving behaviors (i.e., errors, lapses, and traffic violations), and accident risk in the last 12 months. Structural equation modeling analysis revealed that personality traits were associated to aberrant driving behaviors both directly and indirectly. In particular altruism, excitement seeking, and normlessness directly predicted bus drivers’ attitudes toward traffic safety which, in turn, were negatively associated with the three types of self-reported aberrant driving behaviors. Personality traits relevant to emotionality directly predicted bus drivers’ aberrant driving behaviors, without any mediation of attitudes. Finally, only self-reported violations were related to bus drivers’ accident risk. The present findings suggest that the hypothesized personality–attitudes model accounts for aberrant driving behaviors in bus drivers, and provide the empirical basis for evidence-based road safety interventions in the context of public transport. ã 2015 Elsevier Ltd. All rights reserved.
Keywords: Bus drivers Attitudes Errors Lapses Traffic violations Personality Accident risk
1. Introduction Public transportation is a vibrant economic sector in the European Union, and transport through buses comes second to the use of passenger cars (Eurostat, 2014). According to a relevant report from the European Agency for Safety and Health at Work (EU-OSHA), professional drivers face an increased risk for road fatalities. Even light transport vehicles are twice as likely to be involved in crashes, as compared to passenger vehicles, and in most cases (85%), crash involvement is the result of human error (EU-OSHA, 2010). The risk of fatalities or serious injury for bus passengers is considerably lower than that of car passengers (Albertsson and Falkmer, 2005 Yang et al., 2009). In the European
* Corresponding author at: Department of Movement, Human and Health Sciences, University of Rome “Foro Italico”, P.za Lauro de Bosis 15, 00135 Rome, Italy. Tel.: +39 06 36 733 36; Mobile: +39 3204570377. E-mail address: [email protected] (L. Mallia). http://dx.doi.org/10.1016/j.aap.2015.03.034 0001-4575/ ã 2015 Elsevier Ltd. All rights reserved.
Union, bus and coach are the most widespread (12.10%) mode for passenger on land transportation, following car use (81.6%). However, the crashes involving bus and coaches in 2010 accounted only for the 0.36% of the total fatalities in crashes (0.52% for urban areas and 0.47% for rural areas; European Union Road Federation, 2012). Furthermore, the last decades witnessed a significant reduction of fatal crashes in bus/coach transportation, and one of the 10 goals of the European Union is to further improve public transportation and road safety by 2050 (European Union, 2011). To achieve this goal more focused research is needed on the risk factors for bus crashes. One way of looking at this is by attending to technical aspects, such as improving vehicle safety features (e.g., the length of the bus), the roadway (e.g., the presence of bus priority and/or a divided roads), as well as the environment (e.g., the traffic conditions and road congestion; Albertsson and Falkmer, 2005; Chimba et al., 2010; Goh et al., 2014). Another way of looking at improving bus transportation safety concerns driver characteristics and attributes. Specifically, factors
146
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
such as age and gender, work conditions and shifts, experience, celeration behavior and speed choice, as well as sleepiness have been studied with respect to crash risk (e.g., Kaplan and Prato, 2012; Strathman et al., 2010; Tseng, 2012; Vennelle et al., 2010; af Wåhlberg, 2008). A recent study (Goh et al., 2014) analyzing about 7000 Australian bus crashes showed that older age (60 years or older), lower working experience (2 years or less), and being involved in a crash in the past predicted crash involvement in an atfault crash. Nevertheless, studies of traffic accidents and crashes in other professional and non-professional drivers have consistently shown that personality attributes play an important role in predicting crash involvement. In a pioneering early study, Burns and Wilde (1995) showed that a “High Risk Personality” profile (i.e., need for tension, risk and adventure in own lives) was associated with speeding and careless driving among professional taxi drivers, whereas excitement seeking was related to traffic rule violation. More recently, Sümer (2003) showed that, among different groups of professional drivers, depression symptomatology, anxiety, hostility and psychoticism indirectly predicted crashes, through their effect on at wheel violations and errors. Furthermore, the study showed that the excitement seeking trait had a direct effect on speeding, and aggressiveness trait directly related to drunk driving. In addition, a growing body of research in the last decade has emphasized the role of personality characteristics on risky driving and crash risk. Some studies have focused on the impact of single personality dimensions upon risky driving behavior (e.g., Dahlen et al., 2005; Jonah et al., 2001; Lajunen, 2001; Özkan and Lajunen, 2005; Taubman-Ben-Ari and Yehiel, 2012), while others estimated the risk for traffic crashes on the basis of the multivariate combination of different personality dimensions (e.g., Deery and Fildes, 1999; Ulleberg, 2001; Lucidi et al., 2010). Recently some studies used this multivariate perspective to understand the role of personality traits on the driving behavior of drivers of different age groups (e.g., Machin and Sankey 2008; Lucidi et al., 2014; Ulleberg and Rundmo, 2003; Yang et al., 2013). Nevertheless, personality is relatively a stable human characteristic that is not easily malleable by road safety interventions. Furthermore, personality is considered to be a distal predictor of behavior, as compared to other more immediate antecedents of behavioral intention and action initiation (Fishbein and Cappella, 2006). In line with this argument, Nordfjærn et al. (2010) found that, in a representative sample of Norwegian drivers, personality traits, such as anxiety, sensation seeking and normlessness were weakly associated with risky driving, and argued that the personality-risky driving relationship could be better understood after considering more immediate antecedents of driving behavior, such as attitudes towards risky driving. In the behavioral science literature, theoretical models such as the theory of planned behavior and the integrative model posit that the effects of personality traits on both intentions and actual behavioral enactment can be mediated by social cognitive variables, such as attitudes (Ajzen, 2011; Fishbein, 2009). The relationship between personality–attitudes–behavior has been assessed in the driving literature through the Ulleberg and Rundmo (2003) model. This model purports that a group of personality traits is relevant to risky driving, and these traits could affect risky driving tendencies both directly, and indirectly, through the effects of attitudes towards traffic safety. Ulleberg and Rundmo (2003) showed that, among young drivers, some personality traits (i.e., anxiety, hostility, normlessness, excitementseeking, and aggression) were indirectly associated to risky driving, while others (i.e., altruism) were directly associated with risky driving. A recent study confirmed this model in a sample of older Italian drivers (Lucidi et al., 2014), and showed that anxiety
(positively), and hostility and normlessness (negatively) were associated to more positive attitudes towards traffic safety, which, in turn, were directly associated with different types of aberrant behaviors at wheel such as violations (e.g., conscious deviations from rules or safe practices), lapses (e.g., mistakes due to attention and memory failures), and errors (e.g., mistakes due to the failure of a planned action). Following an analysis of the direct effects of personality on these self-reported driving behaviors, the same study found that excitement seeking was directly (positively) associated with violations, while hostility was directly associated with both lapses and errors. Taken together, the aforementioned findings show that Ulleberg and Rundmo (2003) ‘personality-attitudes-risky driving behavior’ model can be applied in different age groups across countries, and validly predict risky driving. However, to the best of our knowledge, no studies have assessed this model among professional groups, such as bus drivers. This leaves an important gap in the driving literature for the following reasons. At a theoretical level, it is important to empirically assess the generalizability and applicability of the Ulleberg and Rundmo (2003) model by extending it to a group of professional drivers. To this end, it is important to assess if the model’s components (e.g., personality traits) differentially predict bus drivers’ risky driving, as compared to previous studies with non-professional drivers. At a practical level, better understanding the influence of personality attributes on bus drivers’ risky driving tendencies will provide valuable input for road safety interventions in this target group. In fact, a study applying the Ulleberg and Rundmo (2003) model to bus drivers can provide the empirical basis and inputs for the design and implementation of practical road safety interventions in line with the EU’s 2050 goals for improving public transportation safety and reducing traffic accidents. In view of these arguments, the present study aims to empirically examine the model of Ulleberg and Rundmo (2003) in a large sample of bus drivers. It is noteworthy, that this is the first time that the specific model is examined in this professional group. It was expected that personality traits would predict aberrant selfreported driving behaviors (driving violations, lapses, and errors) both directly, and through the effects of attitudes towards safe driving. So, in line with Ulleberg and Rundmo (2003), we tested a model that assessed both direct and indirect effects of personality traits on aberrant driving behaviors. A secondary aim of the study was to assess the relationship between the three self-reported behavioral indices (i.e., violations, errors and lapses) with the selfreported crash/near crash involvement. This would add greater external validity to the model and allow us to extend the applicability of the model with respect to observed number of crashes or near crash involvements in bus drivers. 2. Methods 2.1. Sample and procedure Three hundred and one male bus drivers aged between 22 and 60 years (mean = 39.1, SD = 10.7 years) participated in the study. Participants were recruited through a convenience sampling procedure in public transport companies of two big cities of Central and Southern Italy (i.e., Florence and Naples) by trained research associates from Sapienza University of Rome. The participants held the driver’s license from about 15 years, worked as bus drivers for about 12 years and worked mainly within an urban area. The study was approved by the Institutional Review Board of the Department of Social and Developmental Psychology of the University. In line with standards for ethics in behavioral research (e.g., The British Psychological Society’s Code of Ethics), all participants were informed about the study and required to give
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
147
2.2. Assessment
horn to indicate your annoyance to another road user”), errors (8 items, e.g., “Underestimate the speed of an oncoming vehicle when overtaking” and “Fail to check your rear-view mirror before pulling out, changing lanes, etc”) and lapses (8 items, e.g., “Misread the signs and exit from a roundabout on the wrong road” and “Hit something when reversing that you had not previously seen”).
The Italian versions of the following measures were used. These measures have been successfully used in previous studies with Italian drivers of different ages (i.e., Lucidi et al., 2010, 2014).
2.2.4. Crash and near-crash involvement Bus drivers were also asked to indicate if, in the last year, they were involved in crashes and/or in near-crashes as drivers (Yes/No).
2.2.1. Personality measures Personality dimensions were assessed using four facets of the “NEO-Personality Inventory-Revised” (Costa and McCrae, 1992): excitement-seeking (E5) (e.g., “I often crave excitement”), angry hostility (N2) (e.g., “I often get angry at the way people treat me”), anxiety (N1) (e.g., “I often feel tense and jittery”), and altruism (A3) (e.g., “I generally try to be thoughtful and considerate”). Each facet consisted of eight items. Normlessness (i.e., the belief that socially unapproved behaviors are required to achieve certain goals) was assessed with Kohn and Schooler (1983) “Normlessness Scale”. This scale included four items (e.g., “If something works,it is less important whether it is right or wrong”). Higher scores in these measures indicated more of each trait.
2.3. Statistical analysis
their consent for participation. All of the contacted participants agreed to take part in the study (100% response rate) and completed structured anonymous questionnaires, which lasted for approximately 40 min.
2.2.2. Attitudes towards traffic safety Attitudes related to driving were measured through the 11-item attitudes scale developed by Iversen and Rundmo (2004) (e.g., “Traffic rules must be respected regardless of road and weather conditions” and “Punishments for speeding should be more restrictive”). A mean score was computed and higher scores indicated more positive attitudes toward traffic safety (i.e., low preferences for risk-taking in traffic). For personality and attitude measures, the responses were given on 5-point Likert scales, from “strongly disagree” (1) to “strongly agree” (5). 2.2.3. Aberrant driving behaviors Bus drivers’ behavior at wheel was measured through the driver behavior questionnaire (DBQ) a scale widely used in past studies in order to examine different types of self-reported aberrant driving behaviors (Lajunen et al., 2004). The DBQ has been used in different studies across countries (e.g., Gras et al., 2006; Özkan et al., 2006; Lucidi et al., 2010, 2014) and among different groups of professional drivers, such as truck/lorry drivers (e.g., Sullman et al., 2002). For the purposes of the present study the 28-item version was used that was developed by Lawton et al. (1997), and comprised eight errors and eight lapses, along with the extended violations scale (eight highway violations and four aggressive violations). Respondents were required to indicate, on a six-point scale from 0 = never to 5 = nearly all the time, how often in the past year they committed specific driving violations (12 items, e.g., “Disregard the speed limit on a residential road” and “Sound your
Structural equation modeling (SEM), through Mplus software (Version 6; Muthén and Muthén, 2010), was used in order to test the hypothesized model. The model’s parameters were estimated using the weighted least squares means and variance adjusted (WLSMV) estimation method. According to standard procedures for SEM analysis, we used an item parceling procedure (Kim and Hagtevt, 2003) in order to calculate the measurement indicators for all key latent variables of the tested model (i.e., anxiety, hostility, excitement seeking, altruism, normlessness, positive attitudes, violations, lapses and errors). Item parceling combines the items of a scale into a smaller set of items in order to reduce the dimensionality and the number of parameters being estimated in the model, resulting in a more parsimonious measurement model and more stable parameter estimates (Little et al., 2002). More specifically, in the present study, the item parcels for each latent variable were created by randomly grouping the items of each scale in three separate item sets (parcels) and by averaging the item scores within each set. A dichotomous variable (yes/no) about crash and near-crash involvement was considered observed measures of accident risk in the tested model. In order to evaluate the SEM results, we considered as good fit model indices a TLI (Tucker–Lewis index) or CFI (comparative fit index) values close to 0.95, a RMSEA (root mean square error of approximation) value below 0.06 (Hu and Bentler, 1999), a WRMR close to 1 and a x2/df ratio below two (Tabachnick and Fidell, 2007). 3. Results The participants responded to all the questions in the questionnaire, and, thus, there were no missing data on the measured variables of the study. Table 1 reports the sociodemographic characteristics of the bus drivers, and their driving habits and experience. With respect to their driving habits, about the 90% of the drivers reported to drive daily and to drive more than 200 km a week. Furthermore, about the 80% of the participants were driving for work exclusively in an urban area. Table 2 reports the descriptive statistics and the reliability coefficients of the measures used in the study. All the measures showed acceptable internal consistency reliability (i.e., Cronbach alpha >.60) with the anxiety measure. With respect to the
Table 1 Bus drivers’ characteristics, driving habits and experience (N = 301). Mean (SD)/percentage Mean age (SD) Mean years that they have driver’s license (SD) Mean years that they work as bus driver (SD) Driving every-day Driving more than 200 km a week Driving exclusively in urban area for work
39.1 (10.4) 15.3 (11.0) 12.5 (10.4) 91.0% 87.9% 80.0%
Involved at least in a crash/near-missing crash in the last year (self-report)
40.2%
148
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
Table 2 Descriptive statistics and Cronbach’s alpha of the key measures of the study. Response range
Mean (SD)
Cronbach’s alpha
Personality dimensions Anxiety Angry hostility Excitement seeking Altruism Normlessness
1–5 1–5 1–5 1–5 1–5
2.6 (.52) 2.38 (.59) 2.75 (.65) 4.04 (.47) 2.37(.52)
.46 .63 .60 .61 .72
.21 .49 .13 .29 .51
.01 .71 .01 .36 .36
Positive attitude toward traffic safety
1–5
4.24 (.52)
.75
1.30
3.76
Self-reported aberrant driving behaviors Violations Lapses Errors
0–5 0–5 0–5
.85 (.59) .63 (.56) .46 (.52)
.85 .82 .87
2.14 1.80 2.58
3.75 4.40 9.9
Skewness
Kurtosis
Fig. 1. Results and latent effects from SEM analysis about the model hypothesizing that personality traits would predict self-reported driving behavior (driving violations, lapses, and errors) both directly, and through the effects of attitudes towards traffic safety. The continuous unidirectional line in the figure denotes only the significant paths obtained through SEM significant for a p-level = .05. For simplicity reasons, all the notsignificant paths between personality dimensions, attitudes and aberrant driving behaviors, and the paths between aberrant driving behaviors and crash risk were omitted.
responses distribution, the personality dimensions showed a normal distribution (i.e., skewness and kurtosis ranging from 1 to 1), while the attitudes and the self-reported driving behaviors (i.e., violations, lapses and errors) did not. The results of the SEM revealed adequate fit indices for the tested model (x2/df = 1.62; CFI = 0.90; RMSEA = 0.046; WRMR = 1.00). All factor loadings of each latent variable were statistically significant (p < 0.001) and were above 0.32, that is the minimum value that the literature generally suggests to accept for a factor loading (Tabachnick and Fidell, 2007). Fig. 1 presents the paths between the latent variables, while the covariance matrices between the latent variables are reported in Table 3a and b. With respect to the paths between the latent variables (Fig. 1), the results of the SEM analysis showed that, in bus drivers, positive attitudes toward traffic safety were positively associated with scores in altruism (b = .32) and negatively associated with scores in excitement seeking (b = .16) and in normlessness (b = .43). These three personality traits explained an overall 35% of the variance in attitudes. In turn, positive attitudes towards traffic safety were negatively related to self-reported violations (b = .63), lapses (b = .19), and errors (b = .36). Direct effects of personality on self-reported aberrant driving behaviors were
also observed. In particular, hostility scores were positively associated with both violations (b = .28) and errors (b = .17) at the wheel, whereas higher anxiety scores were positively related to self-reported lapses (b = .21). Finally, regarding the relationship of self-reported behaviors at wheel with accident risk measured as Table 3 Covariance matrices between personality dimensions (3a) and between aberrant driving behaviors (3b) derived by SEM analysis. 1
2
3a 1. Anxiety 2. Hostility 3. Excitement seeking 4. Altruism 5. Normlessness
– .60** .03 .17* .03
– .11 .60** .21*
3b 1. Violations 2. Lapses 3. Errors
– .68** .68**
– .85**
* **
The covariance is statistical significant for p-level
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Crash risk and aberrant driving behaviors among bus drivers: The role of personality and attitudes towards traffic safety Luca Mallia a,b, * , Lambros Lazuras a , Cristiano Violani c , Fabio Lucidi a a b c
Department of Psychology of Development and Socialization Processes, Sapienza University of Rome, Via Dei Marsi 78, 00185 Rome, Italy Department of Movement, Human and Health Sciences, University of Rome “Foro Italico”, P.za Lauro de Bosis 15, 00135 Rome, Italy Department of Psychology, Sapienza University of Rome, Via Dei Marsi 78, 00185, Rome, Italy
A R T I C L E I N F O
A B S T R A H A C T
Article history: Received 3 December 2014 Received in revised form 25 February 2015 Accepted 23 March 2015 Available online xxx
Several studies have shown that personality traits and attitudes toward traffic safety predict aberrant driving behaviors and crash involvement. However, this process has not been adequately investigated in professional drivers, such as bus drivers. The present study used a personality–attitudes model to assess whether personality traits predicted aberrant self-reported driving behaviors (driving violations, lapses, and errors) both directly and indirectly, through the effects of attitudes towards traffic safety in a large sample of bus drivers. Additionally, the relationship between aberrant self-reported driving behaviors and crash risk was also assessed. Three hundred and one bus drivers (mean age = 39.1, SD = 10.7 years) completed a structured and anonymous questionnaire measuring personality traits, attitudes toward traffic safety, self-reported aberrant driving behaviors (i.e., errors, lapses, and traffic violations), and accident risk in the last 12 months. Structural equation modeling analysis revealed that personality traits were associated to aberrant driving behaviors both directly and indirectly. In particular altruism, excitement seeking, and normlessness directly predicted bus drivers’ attitudes toward traffic safety which, in turn, were negatively associated with the three types of self-reported aberrant driving behaviors. Personality traits relevant to emotionality directly predicted bus drivers’ aberrant driving behaviors, without any mediation of attitudes. Finally, only self-reported violations were related to bus drivers’ accident risk. The present findings suggest that the hypothesized personality–attitudes model accounts for aberrant driving behaviors in bus drivers, and provide the empirical basis for evidence-based road safety interventions in the context of public transport. ã 2015 Elsevier Ltd. All rights reserved.
Keywords: Bus drivers Attitudes Errors Lapses Traffic violations Personality Accident risk
1. Introduction Public transportation is a vibrant economic sector in the European Union, and transport through buses comes second to the use of passenger cars (Eurostat, 2014). According to a relevant report from the European Agency for Safety and Health at Work (EU-OSHA), professional drivers face an increased risk for road fatalities. Even light transport vehicles are twice as likely to be involved in crashes, as compared to passenger vehicles, and in most cases (85%), crash involvement is the result of human error (EU-OSHA, 2010). The risk of fatalities or serious injury for bus passengers is considerably lower than that of car passengers (Albertsson and Falkmer, 2005 Yang et al., 2009). In the European
* Corresponding author at: Department of Movement, Human and Health Sciences, University of Rome “Foro Italico”, P.za Lauro de Bosis 15, 00135 Rome, Italy. Tel.: +39 06 36 733 36; Mobile: +39 3204570377. E-mail address: [email protected] (L. Mallia). http://dx.doi.org/10.1016/j.aap.2015.03.034 0001-4575/ ã 2015 Elsevier Ltd. All rights reserved.
Union, bus and coach are the most widespread (12.10%) mode for passenger on land transportation, following car use (81.6%). However, the crashes involving bus and coaches in 2010 accounted only for the 0.36% of the total fatalities in crashes (0.52% for urban areas and 0.47% for rural areas; European Union Road Federation, 2012). Furthermore, the last decades witnessed a significant reduction of fatal crashes in bus/coach transportation, and one of the 10 goals of the European Union is to further improve public transportation and road safety by 2050 (European Union, 2011). To achieve this goal more focused research is needed on the risk factors for bus crashes. One way of looking at this is by attending to technical aspects, such as improving vehicle safety features (e.g., the length of the bus), the roadway (e.g., the presence of bus priority and/or a divided roads), as well as the environment (e.g., the traffic conditions and road congestion; Albertsson and Falkmer, 2005; Chimba et al., 2010; Goh et al., 2014). Another way of looking at improving bus transportation safety concerns driver characteristics and attributes. Specifically, factors
146
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
such as age and gender, work conditions and shifts, experience, celeration behavior and speed choice, as well as sleepiness have been studied with respect to crash risk (e.g., Kaplan and Prato, 2012; Strathman et al., 2010; Tseng, 2012; Vennelle et al., 2010; af Wåhlberg, 2008). A recent study (Goh et al., 2014) analyzing about 7000 Australian bus crashes showed that older age (60 years or older), lower working experience (2 years or less), and being involved in a crash in the past predicted crash involvement in an atfault crash. Nevertheless, studies of traffic accidents and crashes in other professional and non-professional drivers have consistently shown that personality attributes play an important role in predicting crash involvement. In a pioneering early study, Burns and Wilde (1995) showed that a “High Risk Personality” profile (i.e., need for tension, risk and adventure in own lives) was associated with speeding and careless driving among professional taxi drivers, whereas excitement seeking was related to traffic rule violation. More recently, Sümer (2003) showed that, among different groups of professional drivers, depression symptomatology, anxiety, hostility and psychoticism indirectly predicted crashes, through their effect on at wheel violations and errors. Furthermore, the study showed that the excitement seeking trait had a direct effect on speeding, and aggressiveness trait directly related to drunk driving. In addition, a growing body of research in the last decade has emphasized the role of personality characteristics on risky driving and crash risk. Some studies have focused on the impact of single personality dimensions upon risky driving behavior (e.g., Dahlen et al., 2005; Jonah et al., 2001; Lajunen, 2001; Özkan and Lajunen, 2005; Taubman-Ben-Ari and Yehiel, 2012), while others estimated the risk for traffic crashes on the basis of the multivariate combination of different personality dimensions (e.g., Deery and Fildes, 1999; Ulleberg, 2001; Lucidi et al., 2010). Recently some studies used this multivariate perspective to understand the role of personality traits on the driving behavior of drivers of different age groups (e.g., Machin and Sankey 2008; Lucidi et al., 2014; Ulleberg and Rundmo, 2003; Yang et al., 2013). Nevertheless, personality is relatively a stable human characteristic that is not easily malleable by road safety interventions. Furthermore, personality is considered to be a distal predictor of behavior, as compared to other more immediate antecedents of behavioral intention and action initiation (Fishbein and Cappella, 2006). In line with this argument, Nordfjærn et al. (2010) found that, in a representative sample of Norwegian drivers, personality traits, such as anxiety, sensation seeking and normlessness were weakly associated with risky driving, and argued that the personality-risky driving relationship could be better understood after considering more immediate antecedents of driving behavior, such as attitudes towards risky driving. In the behavioral science literature, theoretical models such as the theory of planned behavior and the integrative model posit that the effects of personality traits on both intentions and actual behavioral enactment can be mediated by social cognitive variables, such as attitudes (Ajzen, 2011; Fishbein, 2009). The relationship between personality–attitudes–behavior has been assessed in the driving literature through the Ulleberg and Rundmo (2003) model. This model purports that a group of personality traits is relevant to risky driving, and these traits could affect risky driving tendencies both directly, and indirectly, through the effects of attitudes towards traffic safety. Ulleberg and Rundmo (2003) showed that, among young drivers, some personality traits (i.e., anxiety, hostility, normlessness, excitementseeking, and aggression) were indirectly associated to risky driving, while others (i.e., altruism) were directly associated with risky driving. A recent study confirmed this model in a sample of older Italian drivers (Lucidi et al., 2014), and showed that anxiety
(positively), and hostility and normlessness (negatively) were associated to more positive attitudes towards traffic safety, which, in turn, were directly associated with different types of aberrant behaviors at wheel such as violations (e.g., conscious deviations from rules or safe practices), lapses (e.g., mistakes due to attention and memory failures), and errors (e.g., mistakes due to the failure of a planned action). Following an analysis of the direct effects of personality on these self-reported driving behaviors, the same study found that excitement seeking was directly (positively) associated with violations, while hostility was directly associated with both lapses and errors. Taken together, the aforementioned findings show that Ulleberg and Rundmo (2003) ‘personality-attitudes-risky driving behavior’ model can be applied in different age groups across countries, and validly predict risky driving. However, to the best of our knowledge, no studies have assessed this model among professional groups, such as bus drivers. This leaves an important gap in the driving literature for the following reasons. At a theoretical level, it is important to empirically assess the generalizability and applicability of the Ulleberg and Rundmo (2003) model by extending it to a group of professional drivers. To this end, it is important to assess if the model’s components (e.g., personality traits) differentially predict bus drivers’ risky driving, as compared to previous studies with non-professional drivers. At a practical level, better understanding the influence of personality attributes on bus drivers’ risky driving tendencies will provide valuable input for road safety interventions in this target group. In fact, a study applying the Ulleberg and Rundmo (2003) model to bus drivers can provide the empirical basis and inputs for the design and implementation of practical road safety interventions in line with the EU’s 2050 goals for improving public transportation safety and reducing traffic accidents. In view of these arguments, the present study aims to empirically examine the model of Ulleberg and Rundmo (2003) in a large sample of bus drivers. It is noteworthy, that this is the first time that the specific model is examined in this professional group. It was expected that personality traits would predict aberrant selfreported driving behaviors (driving violations, lapses, and errors) both directly, and through the effects of attitudes towards safe driving. So, in line with Ulleberg and Rundmo (2003), we tested a model that assessed both direct and indirect effects of personality traits on aberrant driving behaviors. A secondary aim of the study was to assess the relationship between the three self-reported behavioral indices (i.e., violations, errors and lapses) with the selfreported crash/near crash involvement. This would add greater external validity to the model and allow us to extend the applicability of the model with respect to observed number of crashes or near crash involvements in bus drivers. 2. Methods 2.1. Sample and procedure Three hundred and one male bus drivers aged between 22 and 60 years (mean = 39.1, SD = 10.7 years) participated in the study. Participants were recruited through a convenience sampling procedure in public transport companies of two big cities of Central and Southern Italy (i.e., Florence and Naples) by trained research associates from Sapienza University of Rome. The participants held the driver’s license from about 15 years, worked as bus drivers for about 12 years and worked mainly within an urban area. The study was approved by the Institutional Review Board of the Department of Social and Developmental Psychology of the University. In line with standards for ethics in behavioral research (e.g., The British Psychological Society’s Code of Ethics), all participants were informed about the study and required to give
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
147
2.2. Assessment
horn to indicate your annoyance to another road user”), errors (8 items, e.g., “Underestimate the speed of an oncoming vehicle when overtaking” and “Fail to check your rear-view mirror before pulling out, changing lanes, etc”) and lapses (8 items, e.g., “Misread the signs and exit from a roundabout on the wrong road” and “Hit something when reversing that you had not previously seen”).
The Italian versions of the following measures were used. These measures have been successfully used in previous studies with Italian drivers of different ages (i.e., Lucidi et al., 2010, 2014).
2.2.4. Crash and near-crash involvement Bus drivers were also asked to indicate if, in the last year, they were involved in crashes and/or in near-crashes as drivers (Yes/No).
2.2.1. Personality measures Personality dimensions were assessed using four facets of the “NEO-Personality Inventory-Revised” (Costa and McCrae, 1992): excitement-seeking (E5) (e.g., “I often crave excitement”), angry hostility (N2) (e.g., “I often get angry at the way people treat me”), anxiety (N1) (e.g., “I often feel tense and jittery”), and altruism (A3) (e.g., “I generally try to be thoughtful and considerate”). Each facet consisted of eight items. Normlessness (i.e., the belief that socially unapproved behaviors are required to achieve certain goals) was assessed with Kohn and Schooler (1983) “Normlessness Scale”. This scale included four items (e.g., “If something works,it is less important whether it is right or wrong”). Higher scores in these measures indicated more of each trait.
2.3. Statistical analysis
their consent for participation. All of the contacted participants agreed to take part in the study (100% response rate) and completed structured anonymous questionnaires, which lasted for approximately 40 min.
2.2.2. Attitudes towards traffic safety Attitudes related to driving were measured through the 11-item attitudes scale developed by Iversen and Rundmo (2004) (e.g., “Traffic rules must be respected regardless of road and weather conditions” and “Punishments for speeding should be more restrictive”). A mean score was computed and higher scores indicated more positive attitudes toward traffic safety (i.e., low preferences for risk-taking in traffic). For personality and attitude measures, the responses were given on 5-point Likert scales, from “strongly disagree” (1) to “strongly agree” (5). 2.2.3. Aberrant driving behaviors Bus drivers’ behavior at wheel was measured through the driver behavior questionnaire (DBQ) a scale widely used in past studies in order to examine different types of self-reported aberrant driving behaviors (Lajunen et al., 2004). The DBQ has been used in different studies across countries (e.g., Gras et al., 2006; Özkan et al., 2006; Lucidi et al., 2010, 2014) and among different groups of professional drivers, such as truck/lorry drivers (e.g., Sullman et al., 2002). For the purposes of the present study the 28-item version was used that was developed by Lawton et al. (1997), and comprised eight errors and eight lapses, along with the extended violations scale (eight highway violations and four aggressive violations). Respondents were required to indicate, on a six-point scale from 0 = never to 5 = nearly all the time, how often in the past year they committed specific driving violations (12 items, e.g., “Disregard the speed limit on a residential road” and “Sound your
Structural equation modeling (SEM), through Mplus software (Version 6; Muthén and Muthén, 2010), was used in order to test the hypothesized model. The model’s parameters were estimated using the weighted least squares means and variance adjusted (WLSMV) estimation method. According to standard procedures for SEM analysis, we used an item parceling procedure (Kim and Hagtevt, 2003) in order to calculate the measurement indicators for all key latent variables of the tested model (i.e., anxiety, hostility, excitement seeking, altruism, normlessness, positive attitudes, violations, lapses and errors). Item parceling combines the items of a scale into a smaller set of items in order to reduce the dimensionality and the number of parameters being estimated in the model, resulting in a more parsimonious measurement model and more stable parameter estimates (Little et al., 2002). More specifically, in the present study, the item parcels for each latent variable were created by randomly grouping the items of each scale in three separate item sets (parcels) and by averaging the item scores within each set. A dichotomous variable (yes/no) about crash and near-crash involvement was considered observed measures of accident risk in the tested model. In order to evaluate the SEM results, we considered as good fit model indices a TLI (Tucker–Lewis index) or CFI (comparative fit index) values close to 0.95, a RMSEA (root mean square error of approximation) value below 0.06 (Hu and Bentler, 1999), a WRMR close to 1 and a x2/df ratio below two (Tabachnick and Fidell, 2007). 3. Results The participants responded to all the questions in the questionnaire, and, thus, there were no missing data on the measured variables of the study. Table 1 reports the sociodemographic characteristics of the bus drivers, and their driving habits and experience. With respect to their driving habits, about the 90% of the drivers reported to drive daily and to drive more than 200 km a week. Furthermore, about the 80% of the participants were driving for work exclusively in an urban area. Table 2 reports the descriptive statistics and the reliability coefficients of the measures used in the study. All the measures showed acceptable internal consistency reliability (i.e., Cronbach alpha >.60) with the anxiety measure. With respect to the
Table 1 Bus drivers’ characteristics, driving habits and experience (N = 301). Mean (SD)/percentage Mean age (SD) Mean years that they have driver’s license (SD) Mean years that they work as bus driver (SD) Driving every-day Driving more than 200 km a week Driving exclusively in urban area for work
39.1 (10.4) 15.3 (11.0) 12.5 (10.4) 91.0% 87.9% 80.0%
Involved at least in a crash/near-missing crash in the last year (self-report)
40.2%
148
L. Mallia et al. / Accident Analysis and Prevention 79 (2015) 145–151
Table 2 Descriptive statistics and Cronbach’s alpha of the key measures of the study. Response range
Mean (SD)
Cronbach’s alpha
Personality dimensions Anxiety Angry hostility Excitement seeking Altruism Normlessness
1–5 1–5 1–5 1–5 1–5
2.6 (.52) 2.38 (.59) 2.75 (.65) 4.04 (.47) 2.37(.52)
.46 .63 .60 .61 .72
.21 .49 .13 .29 .51
.01 .71 .01 .36 .36
Positive attitude toward traffic safety
1–5
4.24 (.52)
.75
1.30
3.76
Self-reported aberrant driving behaviors Violations Lapses Errors
0–5 0–5 0–5
.85 (.59) .63 (.56) .46 (.52)
.85 .82 .87
2.14 1.80 2.58
3.75 4.40 9.9
Skewness
Kurtosis
Fig. 1. Results and latent effects from SEM analysis about the model hypothesizing that personality traits would predict self-reported driving behavior (driving violations, lapses, and errors) both directly, and through the effects of attitudes towards traffic safety. The continuous unidirectional line in the figure denotes only the significant paths obtained through SEM significant for a p-level = .05. For simplicity reasons, all the notsignificant paths between personality dimensions, attitudes and aberrant driving behaviors, and the paths between aberrant driving behaviors and crash risk were omitted.
responses distribution, the personality dimensions showed a normal distribution (i.e., skewness and kurtosis ranging from 1 to 1), while the attitudes and the self-reported driving behaviors (i.e., violations, lapses and errors) did not. The results of the SEM revealed adequate fit indices for the tested model (x2/df = 1.62; CFI = 0.90; RMSEA = 0.046; WRMR = 1.00). All factor loadings of each latent variable were statistically significant (p < 0.001) and were above 0.32, that is the minimum value that the literature generally suggests to accept for a factor loading (Tabachnick and Fidell, 2007). Fig. 1 presents the paths between the latent variables, while the covariance matrices between the latent variables are reported in Table 3a and b. With respect to the paths between the latent variables (Fig. 1), the results of the SEM analysis showed that, in bus drivers, positive attitudes toward traffic safety were positively associated with scores in altruism (b = .32) and negatively associated with scores in excitement seeking (b = .16) and in normlessness (b = .43). These three personality traits explained an overall 35% of the variance in attitudes. In turn, positive attitudes towards traffic safety were negatively related to self-reported violations (b = .63), lapses (b = .19), and errors (b = .36). Direct effects of personality on self-reported aberrant driving behaviors were
also observed. In particular, hostility scores were positively associated with both violations (b = .28) and errors (b = .17) at the wheel, whereas higher anxiety scores were positively related to self-reported lapses (b = .21). Finally, regarding the relationship of self-reported behaviors at wheel with accident risk measured as Table 3 Covariance matrices between personality dimensions (3a) and between aberrant driving behaviors (3b) derived by SEM analysis. 1
2
3a 1. Anxiety 2. Hostility 3. Excitement seeking 4. Altruism 5. Normlessness
– .60** .03 .17* .03
– .11 .60** .21*
3b 1. Violations 2. Lapses 3. Errors
– .68** .68**
– .85**
* **
The covariance is statistical significant for p-level
