Transportation Research Part F 65 (2019) 46–55

Contents lists available at ScienceDirect

Transportation Research Part F

journal homepage: www.elsevier.com/locate/trf

The relationship between organizational safety culture and

unsafe behaviors, and accidents among public transport bus
drivers using structural equation modeling
Hamidreza Mokarami a, Seyed Shamseddin Alizadeh b, Tayebe Rahimi Pordanjani c,
Sakineh Varmazyar d,⇑
a
Department of Ergonomics, School of Health, Shiraz University of Medical Sciences, Iran
b
Department of Occupational Health Engineering, Faculty of Health, Tabriz University of Medical Sciences, Iran
c
Department of Psychology, Faculty of Humanities, University of Bojnord, Iran
d
Department of Occupational Health Engineering, Faculty of Health, Qazvin University of Medical Sciences, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Safety culture and unsafe behavior have been considered to be important factors in oper-
Received 13 January 2019 ational safety in various fields, including transportation systems. This study aimed to
Received in revised form 3 July 2019 investigate the relationship between safety culture and unsafe behaviors, and accidents
Accepted 18 July 2019
among bus drivers. Indeed, two questionnaires; i.e., Driver Safety Culture Questionnaire
(DSCQ) and Public Transport Driver Behavior Questionnaire (PTDBQ), were developed.
This cross-sectional study was conducted among 336 public transport bus drivers in
Keywords:
Tehran, Iran. Unsafe behavior and safety culture were measured using DSCQ and PTDBQ,
Transportation
Traffic accident
respectively. Additionally, a questionnaire was developed to assess socio-demographic fac-
Risky behavior tors and number of accidents. DSCQ and PTDBQ had appropriate psychometric properties.
Bus drivers The results indicated a negative relationship between accidents, and safety culture and dri-
Structural Equation Modeling (SEM) vers’ unsafe behaviors. A positive association was also observed between unsafe behaviors
and accidents. On the other hand, the unsafe behavior significantly mediated the relation-
ship between safety culture and accidents. Overall, the model presented in this study indi-
cated that strategies for improvement of organizational safety culture could decrease the
drivers’ unsafe behaviors and subsequently reduce the number of accidents.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction

Public transportation is very important in big cities, such as Tehran, in which a population of about four million is trans-
ported daily (Sahebi, Nassiri, & de Winter, 2019). In some areas, passengers commute in large numbers at the same time. This
makes public transport vital for the vast majority of citizens without access to private transport (Iles, 2005). Most people use
public transportation for work-related purposes. In this regard, this city has one of the highest rates of morbidity and mor-
tality associated with traffic accidents, involving professional and nonprofessional drivers, motorcyclists, and pedestrians
(Attarchi, Dehghan, Seyedmehdi, & Mohammadi, 2012; Mehdizadeh, Shariat-Mohaymany, & Nordfjaern, 2018; Sahebi
et al., 2019).

⇑ Corresponding author.
E-mail address: [email protected] (S. Varmazyar).

https://doi.org/10.1016/j.trf.2019.07.008
1369-8478/Ó 2019 Elsevier Ltd. All rights reserved.
 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55 47

Driver’s behavior is a proximal factor in the road traffic injury causation chain (Mooren, Grzebieta, Williamson, Olivier, &
Friswell, 2014). Drivers’ errors and traffic violations are empirically distinct types of the behavior (Parker, Reason, Manstead,
& Stradling, 1995). Error has been defined as ‘‘the failure of planned actions to achieve intended consequences” (Jahangiri,
Hoboubi, Rostamabadi, Keshavarzi, & Hosseini, 2016; Reason, Manstead, Stradling, Baxter, & Campbell, 1990). On the other
hand, violation (deliberate behavior) has been defined ‘‘as deliberate (though not necessarily reprehensible) deviations from
those practices believed necessary to maintain the safe operation of a potentially hazardous system” (Lawton & Parker,
2002). Drivers’ unsafe behaviors, such as risky violation, highway violation, slip, and lapse, might increase their relative crash
risk (Varmazyar, Mortazavi, Arghami, & Hajizadeh, 2014). The results of the research by Heinrich (1985) showed that more
than 80% of accidents were caused solely by unsafe behaviors. Organizational factors, including driving style, training, and
safety culture, could also be effective in traffic safety (Nævestad, Phillips, & Elvebakk, 2015). Organizational culture affects
employees’ decision-making and behaviors in an organization (Casey, Griffin, Flatau Harrison, & Neal, 2017; Wu, Lin, & Shiau,
2010). This factor reflects shared values (what is important) and beliefs (how things work) that interact with organizational
structures and control systems to produce behavioral norms (the way things are done) (Sharpanskykh & Stroeve, 2011;
Yousefi et al., 2016). Safety culture is a subset of the organizational culture (Cooper Ph. D, 2000) and refers to individuals,
jobs, and organizational characteristics that affect employees’ health and safety (Fernandez-Muniz, Montes-Peon, &
Vazquez-Ordas, 2007; Guldenmund, 2000; Yousefi et al., 2016). This factor also influences the attitudes and safety-related
behaviors of the members of an organization and reduces violations (Cooper Ph. D, 2000; Wu et al., 2010). Organizational
safety culture is an important factor in terms of personnel’s safety (Dorn, 2012; Özkan & Lajunen, 2005). It can enhance
the perception of a safe climate among members and can help predict their perceived safe behaviors (Martínez-Córcoles,
Gracia, Tomás, & Peiró, 2011). Evidence has also indicated that an organization’s safety climate and culture were negatively
associated with unsafe behaviors and accidents (Khosravi et al., 2014).
To the best of our knowledge, the true causal relationship between aberrant behaviors by public transportation drivers
and organizational safety culture has not been investigated in a modeling format. In addition, the role of drivers’ unsafe
behaviors as a mediator in the relationship between safety culture and accidents has not been explored yet. Thus, the present
study contributes to identification of important organizational safety culture factors that affect unsafe behaviors and acci-
dents among bus drivers. It can also help prevent accidents on public transportation systems. This study aims to determine
the causal relationship between organizational safety culture and unsafe behaviors, and accidents among bus drivers using
Structural Equation Modeling (SEM). In this context, two questionnaires; i.e., Driver Safety Culture Questionnaire (DSCQ) and
Public Transport Driver Behavior Questionnaire (PTDBQ), were developed.

2. Materials and methods

2.1. Design and study population

This cross-sectional study was conducted among the public transport bus drivers in Tehran, Iran. The participants were
bus drivers employed by urban public transportation bus company of the city. There were a total of 4332 professional dri-
vers. The inclusion criteria of the study were age < 55 years and job tenure  3 years. The exclusion criteria were having
chronic or acute mental health disorders and/or physical diseases assessed using an author-developed questionnaire. The
495 quilted drivers (age < 55 years and job tenure  3 years) were selected through random proportional sampling (the
number of subpopulations in each area was selected in proportion to the number of population and then, random sampling
techniques were applied to each subpopulation). This was done by assistance of the bus line supervisors of the eight systems
in different sectors of the city according to their personnel codes.
The participants completed DSCQ and PTDBQ and provided demographic and accident information from their work sys-
tems. Out of the 495 bus drivers, 130 did not return the measures or had more than 20% missing data. Besides, 29 drivers
reported having a chronic or acute mental health disorder and/or a physical disease. Therefore, data from 336 participants
(67.88%) were used in the analyses. The 336 self-reported responses were randomly divided into two groups of 168. The
results of independent sample t-test and chi-square test indicated no significant differences between the two groups with
respect to demographic characteristics. The data from the first sample were used for Exploratory Factor Analysis (EFA) of
the questionnaires, while those from the second group were used to for Confirmatory Factor Analysis (CFA) of the question-
naires. Finally, all data from the 336 participants were used to implement SEM.

2.2. Measures/instrument

2.2.1. Determining the dimensions and designs of the questionnaires items

Initial items for DSCQ were generated following a literature review (Chenhall, 2010; Diaz-Cabrera, Hernandez-Fernaud, &
Isla-Diaz, 2007; Fernandez-Muniz et al., 2007). These items were combined with the suggested items from a panel of experts,
including 14 professionals of safety and occupational health employed at the bus company. Totally, an initial 64-item list
was designed. PTDBQ was generated using the Manchester Driver Behavior Questionnaire (DBQ) (af Wåhlberg, Dorn, &
Kline, 2011), field observation of the drivers’ safety behaviors, content analysis of the bus drivers’ violations recorded by Teh-
ran’s traffic police, and unstructured interviews with 42 bus route supervisors and bus drivers. In this way, an initial 46-item
48 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55

list was designed. The accident reports were collected using three self-report questions, including the number of accidents
(damage to the bus), driving offences, and driver at fault in accidents, for the previous three years.
The drivers filled out the DSCQ and PTDBQ using a five-point Likert scale (0 = completely disagree to 4 = completely agree
and 0 = never to 4 = nearly all the time, respectively).

2.2.2. Validity and reliability of the questionnaires

To assess the face validity of the questionnaires, 15 experts reviewed grammar, wording, and item allocation for each
item of the questionnaires. In addition, 20 bus drivers surveyed the questionnaires in order to resolve any ambiguity and
non-understandability. Then, Content Validity Index (CVI) and Content Validity Ratio (CVR) of both questionnaires were
assessed based on the opinions of 15 professionals of occupational health and safety. Relevance of the questionnaires items
was also judged by these experts (according to a 10–point Likert scale). Then, Inter-Class Correlation (ICC) was used to deter-
mine the agreement among the experts to check the relevance of the items.
Cronbach’s a was applied to evaluate the internal consistency of all items in the questionnaires. To determine the test–
retest reliability, 35 subjects were asked to fill out the questionnaires for the second time about two months later. The
repeatability was calculated using Pearson’s correlation coefficient to determine the correlation between the retest and
the initial study (Maasoumi et al., 2017).
Construct validity of the questionnaires was tested using EFA and CFA. Kaiser-Meyer-Olkin (KMO) and Bartlett tests were
performed before factor analysis in order to determine the proportionate data for the EFA. A high KMO value (0.5–1.0) is
appropriate for factor analysis. Indeed, the Bartlett test value should be large enough to be significant. In factor selection,
an eigenvalue of greater than one and the scree plot test were used and a minimum factor loading of 0.40 per item was con-
sidered (Costello & Osborne, 2005; Maasoumi et al., 2017). Principal Component Analysis (PCA) with a varimax rotation was
used to identify the number and nature of the factors for DSCQ and PTDBQ. Then, CFA was performed using the weighted
least squares according to the extracted dimensions of the EFA. The fit indices of v2/df ratio, Comparative Fit Index (CFI),
Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR) were used to con-
firm the measurement models.

2.3. Statistical analysis

SEM invokes a measurement model that defines latent variables using one or more observed variables as well as a struc-
tural model that implies relationships between latent variables. In this method, latent variables that cannot be directly
observed and measured using statistical data are obtained or inferred from a set of observed variables (Hsu et al., 2012). Each
SEM is normally composed of two sub-models: first, a structural model that defines the pattern of relations between the
unobserved factors (endogenous and exogenous variables) and is typically identified in schematic diagrams by the presence
of interrelated ellipses and second, a measurement model or CFA that ‘‘describes the relationship between the latent
variables and the observed variables” (dell’Olio, Ibeas, Oña, & Oña, 2018). In this study, SEM was used to explore causal
relationships between the latent variables of safety culture and unsafe behavior (as a mediator), and accidents, which were
explained by observed variables. Evaluation of the model fit is generally performed based on an inferential goodness-of-fit
index as well as a number of other descriptive and alternative indices. The summary of the fit indices and their acceptable
values has been presented in Table 1.
The statistical analyses were done using IBM SPSS 16 (USA, SPSS Inc.) and LISREL 8.80 (Scientific Software International,
Inc., Lincolnwood, IL, USA).

2.4. Ethics

The drivers were informed about the objectives of the project and that their participation was voluntary. Then, they read
and signed written informed consent forms. Indeed, anonymous questionnaires were used. The study procedures was
reviewed and approved by the ethics committees of Qazvin University of Medical Sciences.

3. Results

The means and standard deviations of the demographic and occupational variables among the bus drivers have been
shown in Table 2. Accordingly, the mean age the drivers was 42.02 ± 5.25 years, ranging from 30 to 54 years. All participants
were male and nearly 98% of them were married.

3.1. Validity and reliability of the questionnaires

Based on the results of CVI and CVR, 22 items from DSCQ and 16 items from PTDBQ were removed. The mean CVIs for
DSCQ and PTDBQ were 0.92 and 0.87, respectively. In addition, the mean CVRs for DSCQ and PTDBQ were 0.76 and 0.70,
respectively. The revised versions with 42 items in DSCQ and 30 items in PTDBQ were used in the next stage. The ICCs were
0.89 and 0.85 for DSCQ and PTDBQ, respectively, which represented excellent consistency and the raters’ agreement on the
 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55 49

Table 1
Descriptions and thresholds of goodness-of-fit indices (Hsu et al., 2012)

Assessment indicator Fit Description Cut-offs

index
Inferential fit index v2 test Indicates the discrepancy between hypothesized model and data P > 0.05
v2/df Its value is divided by the number of degrees of freedom. v2 test is sensitive to sample <2 or <3
size.
Descriptive fit index GFI Comparison of the squared residuals from prediction with the actual data, similar to R2 >0.9 or >0.8
in a regression analysis.
AGFI GFI adjusted for the degrees of freedom, similar to the adjusted R2 used in a regression >0.9 or >0.8
analysis.
NFI Computed by relating the difference of the v2 value for the proposed model from the v2 >0.90
value for the independent or null model.
NNFI NNFI is a simple variant of the NFI that takes into account the degrees of freedom of the >0.90
proposed model.
Alternative fit index NCP Basically reflects the extent to which a proposed model does not fit the data. Close to 0
RMSEA Shows how well the model fits the population covariance matrix. RMSEA is not sample- 0.05: good fit; <0.08:
dependent (unlike the v2 value). reasonable fit
CFI Logic of comparing a proposed model to the null model, assuming no relationships >0.90
between the measures.
AIC AIC value of the theoretical model should be smaller than that of the saturated model –
and the independent model.
SRMR A statistic related to the correlation residuals. <0.05
PGFI >0.50
Measurement model Relationship between the observed and latent variables. Standardized path
assessment coefficients >0.7
Structural model Causal path relationship between the latent variables. R2 > 0.3
assessment

Note: GFI, goodness of fit index; AGFI, adjusted goodness-of-fit index; NFI, normed fit index; NNFI, non-normed fit index; NCP, non-centrality parameter;
RMSEA, root mean square error of approximation; CFI, comparative fit index; AIC, Akaike information criterion; SRMR, standardized root mean square
residual; PGFI, parsimony goodness-of-fit index.

Table 2
Means and standard deviations of the demographic and occupational variables among the
bus drivers (n = 336).

Variable Mean ± SD
Age (years) 42.02 ± 5.25
Job tenure (years) 13.15 ± 3.71
Total driving time per week (h) 58.4 ± 5.48
Distance traveled during the year (km) 38507 ± 9420
Number of accidents (damaged bus) 1.56 ± 1.69
Offences 0.36 ± 1.08
The driver at fault in the previous three years 1.8 ± 1.2

items’ relevance. Moreover, Cronbach’s a coefficients were 0.95 and 0.91 and Pearson’s correlation values were 0.87 and 0.86
for the items of DSCQ and PTDBQ, respectively.
The KMO values were 0.907 and 0.874 for DSCQ and PTDBQ, respectively. In addition, the results of the Bartlett test
(DSCQ: approx. v2 = 3700, df = 861, p < 0.001; and PTDBQ: approx. v2 = 1970, df = 435, p < 0.001) were large and significant.
Therefore, the data were suitable for factor analysis. The results of EFA of DSCQ identified five factors with eigenvalues over
Kaiser’s criterion of 1. These factors were named according to the content analysis of their items as 1) management commit-
ment and safety policy, 2) leadership style and company values, 3) usage of accident information and preventive planning, 4)
participation or driver involvement, and 5) control. These five factors explained 51.7% of the total variance. However, 13
items were removed because they focused on more than one factor or did not load on one factor. Ultimately, 29 items
remained (Table 3). More details regarding the development of this questionnaire have been provided elsewhere by the
authors (Varmazyar, Mortazavi, Arghami, & Hajizadeh, 2016).
The results of EFA of PTDBQ identified five factors with eigenvalues over Kaiser’s criterion of 1. These factors were named
according to the content analysis of their items as 1) highway violations, 2) dangerous violations, 3) risky violations, 4) ordi-
nary violations, and 5) moving violations. These five factors explained 51.6% of the total variance. However, four items were
removed because they focused on more than one factor or did not load on one factor. Ultimately, 26 items remained
(Table 4).
Similarly, the CFA results confirmed the five-factor DSCQ and the five-factor PTDBQ. The standard loads (values to the
right of the brackets), errors (numbers to the left of parentheses), and t-values (numbers in parentheses) of each CFA for
DSCQ and PTDBQ have been presented in Figs. 1 and 2. The bi-directional curved arrows indicate exploratory correlations
50 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55

Table 3
The results of factor analysis, mean, SD, and Cronbach’s a of DSCQ.

Factor Item (factor loading) Variance Communality Corrected item- Cronbach’s

explained total correlation a
Leadership style and company 1. Sincerity and participation of drivers 34.6 0.695 0.605 0.90
values (0.69)
2. Safe transport objectives and programs 0.671 0.584
(0.67)
3. Information about safety rules (0.66) 0.679 0.591
4. How to prevent hazards (0.65) 0.664 0.576
5. Encouragement of drivers to improve 0.709 0.618
transport safety (0.55)
6. Identification of solutions for working 0.603 0.526
errors (0.49)
7. Driver’s initiative in finding new 0.675 0.587
solutions (0.48)
8. Safety instructions for working with the 0.726 0.639
vehicle (0.48)
9. Objectives and benefits of safe transport 0.693 0.602
(0.48)
10. Driver’s collaboration and participation 0.726 0.640
in promoting safety (0.41)
Use of accident information and 11. Training courses for safety equipment 5.35 0.726 0.640 0.86
prevention programs (0.70)
12. Accident prevention programs for 0.717 0.625
anticipated hazards (0.67)
13. Sharing accident prevention programs 0.624 0.544
with drivers (0.65)
14. Use of information from accident 0.733 0.649
investigations (0.50)
15. Consequence of breaking the rules (0.48) 0.660 0.566
16. Training courses about work-related 0.688 0.597
health issues (0.44)
17. Determining training programs (0.44) 0.590 0.518
Management commitment and 18. Driver safety in workplace (0.68) 4.56 0.602 0.525 0.81
safety policy 19. Financial and human resources (0.66) 0.694 0.603
20. Meetings about current safety programs 0.698 0.608
(0.62)
21. Attention to vehicle safety (0.60) 0.659 0.565
22. Special operational policies and 0.729 0.642
procedures (0.48)
23. Satisfaction with transport safety (0.41) 0.631 0.551
Participation 24. Creation of instructions for safe 3.93 0.572 0.501 0.77
transport (0.57)
25. Meetings between officials and drivers 0.628 0.548
(0.54)
26. Employment of experienced drivers 0.580 0.509
(0.53)
Control 27. Recording and analysis of accidents 3.22 0.599 0.523 0.71
(0.71)
28. Previous successful programs (0.41) 0.641 0.558
29. Technical and safety inspections (0.46) 0.683 0.594

Note: DSCQ, driver safety culture questionnaire.

between the measurement errors of the variables. Furthermore, the evaluated fit indices suggested an acceptable goodness-
of-fit for the two questionnaires (Table 5).

3.2. SEM

In order to explore the relationship between safety culture and unsafe behavior (as the mediator), and accidents among
the drivers, SEM analysis was used. The results indicated a negative relationship between the safety culture and the drivers’
unsafe behaviors (b = 0.42, p < 0.001) and accidents (b = 0.27, p < 0.001). In addition, there was a positive association
between the unsafe behaviors and accidents (b = 0.32, p < 0.001). The unsafe behavior significantly mediated the relationship
between the safety culture and accidents. The influence of the safety culture (exogenous variables) on unsafe behavior and
accidents (endogenous variables) with standard loads and errors has been depicted in Fig. 3. The t-values showed that all
factor loadings were significant at 95% Confidence Interval (CI; t-values greater than 1.96). Moreover, the evaluated fit
indices for the final model suggested an acceptable goodness-of-fit for the data (Table 6).
 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55 51

Table 4
The results of factor analysis, mean, SD, and Cronbach’s a of PTDBQ.

Factor Item (factor loadings) Variance Communality Corrected item-total Cronbach’s

explained correlation a
Highway 1. Speeding to avoid a red light (0.68) 29.63 0.667 0.671 0.87
violations 2. Overtaking a slow-moving vehicle (0.66) 0.683 0.678
3. Overtaking in dangerous conditions (0.65) 0.606 0.601
4. Speeding to go through a yellow light (0.65) 0.651 0.659
5. Chasing while angry (0.64) 0.551 0.549
6. Moving while passengers are boarding (0.61) 0.730 0.733
7. Crossing a solid line (0.60) 0.577 0.574
8. Running a red light late at night (0.54) 0.553 0.551
9. Disregarding speed limit late at night (0.52) 0.588 0.585
Dangerous 10. Talking to colleagues or passengers (0.74) 7.17 0.639 0.640 0.83
violations 11. Using a mobile phone (0.73) 0.611 0.604
12. Eating and drinking while driving (0.72) 0.589 0.585
13. Disregarding the speed limit (0.68) 0.547 0.545
14. Arguing with a passenger (0.58) 0.638 0.630
Risky violations 15. Stopping outside a bus stop (0.69) 5.64 0.595 0.591 0.72
16. Moving before closing the bus doors (0.64) 0.569 0.566
17. Colleague riding on the first step of the bus 0.580 0.578
(0.60)
18. Boarding passengers outside the bus stop (0.60) 0.586 0.585
Ordinary 19. Unloading passengers at a red light (0.68) 4.69 0.639 0.640 0.66
violations 20. Suddenly changing lanes (0.59) 0.629 0.622
21. Braking suddenly (0.54) 0.597 0.593
22. Crossing intersection before the light turns 0.568 0.565
green (0.49)
Moving 23. Not observing the safe distance from the front 4.47 0.591 0.588 0.60
violations vehicle (0.69)
24. Not observing pedestrian right of way (0.65) 0.687 0.682
25. Blocking the pedestrian lane (0.51) 0.663 0.665
26. Avoiding an oncoming vehicle (0.41) 0.778 0.785

Note: PTDBQ, public transport driver behavior questionnaire.

4. Discussion

In this study, the organizational safety culture and unsafe behavior factors were identified among Iranian public transport
bus drivers and two new valid questionnaires were developed. The influence of safety culture on accidents and the mediator
role of drivers’ unsafe behaviors were also tested using SEM.

4.1. The psychometric properties of DSCQ and PTDBQ

The psychometric properties of DSCQ and PTDBQ indicated that the five-factor structures of both questionnaires had the
potential for usability in studies among bus drivers of a public transport company in Iran.
The first factor of DSCQ had the highest mean, which indicated relatively high agreement among the drivers on the items
about the behavioral style of the immediate superior, participation in safety issues, providing new solutions, and improving
safety. The second and third factors were composed of items related to programs for prevention of accidents, usage of acci-
dent information, management commitment, and safety policies. Similar results from other studies indicated a major role for
managers in promotion of employees’ safe behaviors (Fernandez-Muniz et al., 2007). These three factors were consistent
with the results of the studies conducted by Diaz-Cabrera et al. (2007) and Chenhall (2010), which revealed that items, such
as initiating solutions to improve safety policies, sincere participation in safety promotion, identification of training needs,
and information about the consequences of non-observance of rules, were loaded on the same factors by employees of dif-
ferent companies. The fourth and fifth factors included the use of drivers’ experiences, joint meetings between officials and
drivers, survey of previous successful programs, and technical and safety inspections. These comprised the drivers’ percep-
tions of participation for improving working conditions and achieving organizational goals. These factors were similar to the
variables in the model proposed by Fernandez-Muniz et al. (2007). They disclosed that participation and control were parts
of the safety management system that affected the safe performance of employees in industrial, construction, and service
sectors.
The first factor of PTDBQ mainly consisted of such behaviors as running red lights (Shi, Bai, Ying, & Atchley, 2010) and
disregarding the speed limit late at night as highway violations, which were similar to the results obtained by Davey,
Wishart, Freeman, and Watson (2007). These violations can occur while rushing to compensate for the lack of time allowed
to traverse the number of stops determined by the bus company. The second factor was dangerous violations and included
items, such as use of a mobile phone, speaking with colleagues, and eating and drinking while driving. Previous studies have
also listed similar violations (De Winter & Dodou, 2010). The reasons for such violations include not allowing sufficient time
52 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55

Fig. 1. Standard loads, errors, and t-values of the measurement model of the Driver Safety Culture Questionnaire (DSCQ).

Fig. 2. Standard loads, errors, and t-values of the measurement model of the Public Transport Driver Behavior Questionnaire (PTDBQ).

for breakfast and tea breaks. It is important to consider these issues as a risk of collision. These results were four times higher
when using a mobile phone (Lam, 2002). Indeed, eating and drinking led to inattention. Based on the results, driving inat-
tention was a contributing factor to 78% of all accidents and 65% of all near-accidents (Klauer, 2005). The third factor was
risky violations, which was similar to the factors found in a previous study (Shi et al., 2010). These violations may occur
due to the high traffic volume on the streets of Tehran, the time required for elderly passengers to board and exit the
 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55 53

Table 5
The results of goodness of fit indices for measurement models or confirmatory factor analysis of DSCQ and PTDBQ.

Measurement v2 df v2/df P-value CFI RMSEA SRMR

DSCQ 3.53 3 1.17 0.316 1.00 0.033 0.013
PTDBQ 2.92 4 0.73 0.570 1.00 0.000 0.015

Note: DSCQ, driver safety culture questionnaire; PTDBQ, public transport driver behavior questionnaire; CFI, comparative fit index; RMSEA, root mean
square error of approximation; SRMR, standardized root mean square residual; SRMR, standardized root mean square residual.

χ2 = 60.59; df = 61; P = 0.49076; RMSEA = 0.000

Fig. 3. Structural equation model of safety culture with unsafe behaviors and accidents.

Table 6
The results of goodness of fit indices for the studied model.

v2 test v2/df NCP RMSEA Model AIC < saturated AIC NFI NNF
and independent AIC
P = 0.49 0.99 0.00 0.000 120.59 < 182 and 1609.22 0.96 1.00
CFI SRMR GFI AGFI PGFI Measurement model Structural model
assessment assessment
1.00 0.039 0.95 0.92 0.63 0.43–0.97 0.18 and 0.25

Note: GFI, goodness of fit index; AGFI, adjusted goodness-of-fit index; NFI, normed fit index; NNFI, non-normed fit index; NCP, non-centrality parameter;
RMSEA, root mean square error of approximation; CFI, comparative fit index; AIC, Akaike information criterion; SRMR, standardized root mean square
residual; PGFI, parsimony goodness-of-fit index.

bus, colleagues’ requests to board and exit outside a bus stop, and hurrying to complete the time allowed for the routes as
determined by the bus company. The fourth factor was ordinary violations that were committed by most drivers. These items
included changing lanes suddenly, braking suddenly, allowing passengers to exit at a red light, and entering an intersection
before the traffic light turns green. All these violations are rooted in drivers’ hurry to save time or to reduce the volume of
passengers at the bus stops and non-observance of the bus right of way by other road users that is related to the size of the
bus and their worry about being blocked from entering a route. The same factor was found by Özkan and Lajunen (2005)
among British, Finnish, and Dutch drivers, which included such items as ‘‘crossing a junction knowing that the traffic lights
have already turned against you”. The fifth factor was moving violations and included behaviors, such as non-observance of a
safe distance from the preceding vehicle, avoiding an oncoming vehicle, non-compliance with pedestrian right of way, and
blocking pedestrian lanes. The reasons for such violations were related to pedestrians’ and other road users’ disrespect for
the bus right of way owing to the size of the bus and their worry about being blocked from entering a route. Other factors
include the difference in the location of the doors of single-cabin and double-cabin buses at some bus stops and a bus being
already present at the bus stop, which can block pedestrian lanes. This factor has been discussed in a previous study (De
Winter & Dodou, 2010). The disparity of PTDBQ factors in various studies can be justified by differences in social and cultural
contexts, road traffics, and target groups of professional drivers.
54 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55

In the current study, the standard loads for CFA of DSCQ showed good explanatory correlation between the observed and
latent variables for safety culture. The t-values indicated the acceptable validity of the observed and latent variables of DSCQ.
Indeed, three factors (manager commitment and safety policy, participation, and accident prevention programs and use of
information about accidents) showed maximum correlations with one another. Thus, these factors, from the drivers’ per-
spectives, were important to promotion of a safety culture in the public transportation bus company. Moreover, considering
financial and human resources by addressing vehicle safety and driver safety in the workplace, holding meetings with dri-
vers concerning safety programs, and holding training courses related to health and technical issues increased the satisfac-
tion of the drivers about transportation safety and reduced the number of driving accidents. Other variables that showed
good correlations indicated that officials and occupational health and safety experts of the bus company should be commit-
ted in their attitudes and behaviors about observation of rules, safety principles, and company’s policies and also use acci-
dent analysis and root causes to prevent further accidents. In the same line, Fernandez-Muniz et al. (2007) found that
managers needed to be wholly committed to and personally involved with safety activities.
Similar to the results of a study conducted by Martinussen, Hakamies-Blomqvist, Moller, Ozkan, and Lajunen (2013), the
current study findings showed that standard loads related to the CFA of PTDBQ had relatively high correlations among
observed and latent variables for unsafe behaviors. The t-values of greater than 1.96 indicated the acceptable fit of the model
at the 95% CI. Thus, the observed variables related to PTDBQ showed good validity and fitness indices. The highest correla-
tions were related to highway, ordinary, and risky violations. These violations were related to appropriate scheduling for
traversing routes from the origin to the destination, suitable timing for eating, drinking, and rest stops for drivers, and dri-
vers’ reporting of accidents owing to non-observance of safety principles or disregarding traffic laws. Provision of informa-
tion concerning these offences could reduce such violations.

4.2. The relationship between organizational safety culture and unsafe behaviors, and accidents

Modeling the influence of safety culture on drivers’ unsafe behaviors showed a negative correlation between the two
latent variables. Thus, promotion of a safety management system could improve the employees’ safe behaviors.
Fernandez-Muniz et al. (2007) also observed the important role of managers in the promotion of employees’ safe behaviors
related to their attitudes and behaviors (direct) and development of a safety management system (indirect). Martinussen
et al. (2013) reported similar results, indicating that when the safety culture had strong leadership, the behavior generated
a higher safety climate among members and predicted their perceived safety behaviors. Furthermore, adherence to different
dimensions of a safety culture could enhance safe behaviors or decrease unsafe behaviors among drivers.
The study results revealed a negative relationship between safety culture and accidents. Accordingly, in addition to indi-
rectly influencing a decrease in unsafe behaviors, the safety culture had a direct effect through vehicle safety and safe routes,
which decreased the number of accidents. Özkan and Lajunen (2005) also found the positive impact of organizational safety
culture on reduction of the number of accidents. Moreover, the results of other studies about organizational safety culture
and qualitative analysis of bus drivers’ behaviors at work showed that policies and practices that increased time pressure
increased the risk of accidents (Dorn, 2012).
The present study findings indicated that accidents could be predicted through statistics about unsafe behaviors. Thus,
increase in drivers’ violations would increase the number of accidents because disregarding driving laws would increase
the chance of involvement in accidents. This result was in line with those of other studies conducted in different countries
concerning the influence of errors and violations on involvement in accidents (Bener, Ozkan, & Lajunen, 2008; Davey et al.,
2007; Kontogiannis, Kossiavelou, & Marmaras, 2002; Sullman, Meadows, & Pajo, 2002).
Despite its strengths and use of SEM, this study had some limitations including its cross-sectional design and self-report
measures. The authors did not have access to the accidents records and, as a result, they had to use the drivers’ self-report
data. Additionally, this project was conducted only in one city and included no female participants.

5. Conclusion

This study presented reliable and valid scales for measuring organizational safety culture and drivers’ unsafe behaviors in
a heavy vehicle transportation system. These scales could also be used to assess the effectiveness of safety measures in
reducing drivers’ aberrant behaviors and occupational accidents in the public transportation sector. The results of modeling
indicated the explanatory role of safety culture in promotion of drivers’ safe behaviors and reduction of the accident rate. The
mediator role of unsafe behavior in the relationship between the safety climate and accidents was confirmed, as well. Yet,
further research is suggested to investigate the influence of the dimensions of safety culture on reduction of unsafe behaviors
and accidents.

Acknowledgments

The authors express their gratitude to the senior and middle managers of the Bus Company and HSE experts in Tehran
transportation Bus Company. We are so grateful to Ms. A. Keivanshekouh at the Research Improvement Center of Shiraz
University of Medical Sciences for improving the use of English in the manuscript.
 H. Mokarami et al. / Transportation Research Part F 65 (2019) 46–55 55

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.trf.2019.07.008.

References

af Wåhlberg, A., Dorn, L., & Kline, T. (2011). The Manchester Driver Behaviour Questionnaire as a predictor of road traffic accidents. Theoretical Issues in
Ergonomics Science, 12, 66–86.
Attarchi, M. S., Dehghan, F., Seyedmehdi, S. M., & Mohammadi, S. (2012). Traffic accidents and related injuries in Iranian professional drivers. Journal of Public
Health, 20, 499–503.
Bener, A., Ozkan, T., & Lajunen, T. (2008). The Driver behaviour questionnaire in Arab Gulf countries: Qatar and United Arab Emirates. Accident Analysis and
Prevention, 40, 1411–1417.
Casey, T., Griffin, M. A., Flatau Harrison, H., & Neal, A. (2017). Safety climate and culture: Integrating psychological and systems perspectives. Journal of
Occupational Health Psychology, 22, 341–353.
Chenhall, E. C. (2010). Assessing safety culture, values, practices, and outcomes. Libraries: Colorado State University.
Cooper Ph. D, M. (2000). Towards a model of safety culture. Safety Science, 36, 111–136.
Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical
Assessment, Research & Evaluation, 10, 1–9.
Davey, J., Wishart, D., Freeman, J., & Watson, B. (2007). An application of the driver behaviour questionnaire in an Australian organisational fleet setting.
Transp Res Part F Traffic Psychol Behav, 10, 11–21.
De Winter, J., & Dodou, D. (2010). The driver behaviour questionnaire as a predictor of accidents: A meta-analysis. Journal of Safety Research, 41, 463–470.
dell’Olio, L., Ibeas, A., Oña, J. d., & Oña, R. d. (2018). Chapter 8 - Structural equation models. In L. dell’Olio, A. Ibeas, J. d. Oña, & R. d. Oña (Eds.), Public
transportation quality of service (pp. 141–154). Elsevier.
Diaz-Cabrera, D., Hernandez-Fernaud, E., & Isla-Diaz, R. (2007). An evaluation of a new instrument to measure organisational safety culture values and
practices. Accident Analysis and Prevention, 39, 1202–1211.
Dorn, L. (2012). Investigating safety culture: A qualitative analysis of bus driver behaviour at work. In Contemporary Ergonomics and Human Factors 2012:
Proceedings of the international conference on Ergonomics & Human Factors 2012, Blackpool, UK, 16-19 April 2012 (pp. 313). CRC Press.
Fernandez-Muniz, B., Montes-Peon, J. M., & Vazquez-Ordas, C. J. (2007). Safety culture: Analysis of the causal relationships between its key dimensions.
Journal of Safety Research, 38, 627–641.
Guldenmund, F. W. (2000). The nature of safety culture: A review of theory and research. Safety Science, 34, 215–257.
Heinrich, H. W. (1985). Industrial accident prevention. A scientific approach. New York: McGraw Hill.
Hsu, I.-Y., Su, T.-S., Kao, C.-S., Shu, Y.-L., Lin, P.-R., & Tseng, J.-M. (2012). Analysis of business safety performance by structural equation models. Safety Science,
50, 1–11.
Iles, R. (2005). Public transport in developing countries. London: Elsevier Ltd.
Jahangiri, M., Hoboubi, N., Rostamabadi, A., Keshavarzi, S., & Hosseini, A. A. (2016). Human error analysis in a permit to work system: A case study in a
chemical plant. Safety and Health at Work, 7, 6–11.
Khosravi, Y., Asilian-Mahabadi, H., Hajizadeh, E., Hassanzadeh-Rangi, N., Bastani, H., & Behzadan, A. H. (2014). Factors influencing unsafe behaviors and
accidents on construction sites: A review. International Journal of Occupational Safety and Ergonomics, 20, 111–125.
Klauer, S. G. (2005). Assessing the effects of driving inattention on relative crash risk. Virginia Tech, Blacksburg(VA): Virginia Polytechnic Institute and State
University.
Kontogiannis, T., Kossiavelou, Z., & Marmaras, N. (2002). Self-reports of aberrant behaviour on the roads: Errors and violations in a sample of Greek drivers.
Accident Analysis and Prevention, 34, 381–399.
Lam, L. T. (2002). Distractions and the risk of car crash injury: The effect of drivers’ age. Journal of Safety Research, 33, 411–419.
Lawton, R., & Parker, D. (2002). Barriers to incident reporting in a healthcare system. BMJ Quality & Safety, 11, 15–18.
Maasoumi, R., Mokarami, H., Nazifi, M., Stallones, L., Taban, A., Yazdani Aval, M., & Samimi, K. (2017). Psychometric properties of the Persian translation of
the sexual quality of life-male questionnaire. American Journal of Men’s Health, 11, 564–572.
Martínez-Córcoles, M., Gracia, F., Tomás, I., & Peiró, J. M. (2011). Leadership and employees’ perceived safety behaviours in a nuclear power plant: A
structural equation model. Safety Science, 49, 1118–1129.
Martinussen, L. M., Hakamies-Blomqvist, L., Moller, M., Ozkan, T., & Lajunen, T. (2013). Age, gender, mileage and the DBQ: The validity of the Driver Behavior
Questionnaire in different driver groups. Accident Analysis and Prevention, 52, 228–236.
Mehdizadeh, M., Shariat-Mohaymany, A., & Nordfjaern, T. (2018). Accident involvement among Iranian lorry drivers: Direct and indirect effects of
background variables and aberrant driving behaviour. Transportation Research Part F: Traffic Psychology and Behaviour, 58, 39–55.
Mooren, L., Grzebieta, R., Williamson, A., Olivier, J., & Friswell, R. (2014). Safety management for heavy vehicle transport: A review of the literature. Safety
Science, 62, 79–89.
Nævestad, T.-O., Phillips, R. O., & Elvebakk, B. (2015). Traffic accidents triggered by drivers at work–A survey and analysis of contributing factors.
Transportation Research Part F: Traffic Psychology and Behaviour, 34, 94–107.
Özkan, T., & Lajunen, T. (2005). A new addition to DBQ: Positive driver behaviours scale. Transportation Research Part F: Traffic Psychology and Behaviour, 8,
355–368.
Parker, D., Reason, J. T., Manstead, A. S., & Stradling, S. G. (1995). Driving errors, driving violations and accident involvement. Ergonomics, 38, 1036–1048.
Reason, J., Manstead, A., Stradling, S., Baxter, J., & Campbell, K. (1990). Errors and violations on the roads: A real distinction? Ergonomics, 33, 1315–1332.
Sahebi, S., Nassiri, H., & de Winter, J. C. (2019). Correlates of self-reported driving aberrations in Tehran: A study at the level of drivers and districts.
Transportation Research Part F: Traffic Psychology and Behaviour..
Sharpanskykh, A., & Stroeve, S. H. (2011). An agent-based approach for structured modeling, analysis and improvement of safety culture. Computational and
Mathematical Organization Theory, 17, 77–117.
Shi, J., Bai, Y., Ying, X., & Atchley, P. (2010). Aberrant driving behaviors: A study of drivers in Beijing. Accident Analysis and Prevention, 42, 1031–1040.
Sullman, M. J., Meadows, M. L., & Pajo, K. B. (2002). Aberrant driving behaviours amongst New Zealand truck drivers. Transportation Research Part F: Traffic
Psychology and Behaviour, 5, 217–232.
Varmazyar, S., Mortazavi, S., Arghami, S., & Hajizadeh, E. (2014). Factor analysis of driver behavior questionnaire (DBQ) in public transportation Bus
Company: Investigation of the relationship between DBQ factors and crashes. Scientific Journal of Review, 3, 155–165.
Varmazyar, S., Mortazavi, S. B., Arghami, S., & Hajizadeh, E. (2016). Relationship between organisational safety culture dimensions and crashes. International
Journal of Injury Control and Safety Promotion, 23, 72–78.
Wu, T. C., Lin, C. H., & Shiau, S. Y. (2010). Predicting safety culture: The roles of employer, operations manager and safety professional. Journal of Safety
Research, 41, 423–431.
Yousefi, Y., Jahangiri, M., Choobineh, A., Tabatabaei, H., Keshavarzi, S., Shams, A., & Mohammadi, Y. (2016). Validity assessment of the Persian version of the
Nordic Safety Climate Questionnaire (NOSACQ-50): A case study in a steel company. Safety and Health at Work, 7, 326–330.