Transportation Research Part F 13 (2010) 32–40

Contents lists available at ScienceDirect

Transportation Research Part F

journal homepage: www.elsevier .com/locate / t r f

Professional and non-professional drivers’ stress reactions and risky driving

Bahar Öz a,*, Türker Özkan b, Timo Lajunen b

a University of Helsinki, Department of Psychology, Human Factors and Safety Behavior Group, Finlandb Middle East Technical University, Department of Psychology, Safety Research Unit, Turkey

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

Article history:Received 11 July 2008Received in revised form 28 September2009Accepted 1 October 2009

Keywords:Driver behavioursDriver stressProfessional driversNon-professional driversRisky traffic behavioursSpeedingAccident involvement

1369-8478/$ - see front matter � 2009 Elsevier Ltddoi:10.1016/j.trf.2009.10.001

* Corresponding author. Tel.: +90 312 2103154; fE-mail addresses: bahar.oz@helsinki.fi, ozbahar@

The aim of the present study was to investigate stress reactions, speeding, number of pen-alties and accident involvement among different driver groups (taxi drivers, minibus driv-ers, heavy vehicle drivers, and non-professional drivers). A total number of 234 maledrivers participated in the study. The participants were asked to complete the Driver StressInventory (DSI) together with a demographic information form. Five dimensions of the DSIwere measured; aggression, dislike of driving, hazard monitoring, fatigue proneness, andthrill-seeking. After controlling the effects of age and annual mileage, the results of theANCOVAs revealed differences between different driver groups in terms of both risky driv-ing behaviours and stress reactions in traffic. Regression analyses indicated that aggression,dislike of driving, and hazard monitoring dimensions of the DSI were related to accidentinvolvement after controlling for the effects of age, annual mileage and driver group. Dis-like of driving and thrill-seeking dimensions of the DSI were related to speeding on in-cityroads.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Professional drivers, i.e., people whose job is driving, are at a high risk of road traffic accidents worldwide. For example,about 25% of fatal work-related accidents in the USA (Toscano & Windau, 1994) and more than 20% of fatal work accidents inDenmark, Finland and Sweden are road traffic accidents (Charbotel, Chiron, Martin, & Bergeret, 2001). A vast amount of stud-ies have been conducted to understand the nature of on-the road behaviours and their relationships to traffic accidentsamong non-professional drivers. However, accidents among professional drivers have attracted very little attention amongsafety researchers (Salminen & Lähdeniemi, 2002). The aim of the present study, therefore, was to investigate the differencesamong driver groups (e.g., professional and non-professional) in terms of stress reactions, speeding, number of penalties andaccident involvement and to examine the relationships between driver stress and risky driving.

Dorn and Brown (2003) suggested that professional drivers are at a high risk of being involved in road traffic accidentsdue to their high annual mileage. In addition, as compared to non-professional drivers, professional driving requires differentdemands from drivers. Driving task demands of professional drivers are, for instance, largely pre-determined. However, driv-ing is a more self-paced task for non-professional drivers and they can largely determine the difficulty and risk level of theirdriving (Caird & Kline, 2004). Non-professional drivers can also choose the mode of transportation, time of travel, and targetspeed while driving. Driving is rather a less self-regulated task for professional drivers, because many different factors (e.g.,time schedule, long working hours) increase their task demands. Professional driving has another distinct aspect: manyorganizational factors like a company’s culture, safety policy and practices as well as safety climate largely determinehow safely a professional driver drives. Moreover, a professional driver working in an organization has usually very limited

. All rights reserved.

ax: +90 312 2107975.metu.edu.tr (B. Öz).

B. Öz et al. / Transportation Research Part F 13 (2010) 32–40 33

possibility to influence these organizational factors (Caird & Kline, 2004). All these issues might make the level and thesources of stress different for the professional and non-professional drivers (Dorn & Brown, 2003).

1.1. Driver stress

Numerous studies have been conducted to identify personality factors related to various types of driver stress vulnera-bility (e.g., Matthews, 2001; 2002). Some of the researchers investigating driver stress have employed transactional modelsof stress (e.g., Gulian, Matthews, Glendon, Davies, & Debney, 1989; Matthews, 2001; 2002). Transactional model of stressincludes cognitions, and the ecological relationship between person and environment (Matthews, 2002). As Matthews(2002) emphasized, a transactional model differentiates different classes of constructs interacting dynamically. These factorsinclude personality factors, cognitive factors, environmental factors, and as outcomes subjective stress symptoms and per-formance. The model suggests that environmental (e.g., high workload) and personality factors (e.g., dislike of driving) deter-mine how external factors are interpreted, which in turn influences cognitive stress processes. Cognitive stress processessupport two forms of outcome: subjective outcomes (e.g., anger) and performance outcomes (e.g., risk-taking). Feedbacksfrom outcomes to environment go dynamically. Most of the time, the stressors are corrected over short period of time ifthe cognitive processing is not highly biased. Matthews (2002) emphasized that when cognitive processing is highly biased,stress outcomes might be more damaging for safety.

Glendon et al. (1993) found five distinct but modestly intercorrelated dimensions of driver stress. These dimensions were‘‘dislike of driving”, ‘‘aggression”, ‘‘alertness”, ‘‘irritation when overtaken” and ‘‘overtaking affect”. Studies showed that bothdislike of driving – the dimension which is most strongly related to negative emotional reactions to driving – and aggression– the dimension which is related to feelings of post-drive anger – were related to some emotional and behavioural stressreactions (e.g., Matthews, Dorn, & Glendon, 1991; Matthews & Wells, 1996). Later, some studies using the DBI revealed dif-ferent factor solutions. Lajunen and Summala (1995) found only three factors, namely aggression, dislike of driving and alert-ness. The other two overtaking factors loaded predominantly on aggression. Westerman and Haigney (2000) suggested twonew ‘situation-specific’ factors in a five-factor solution. These factors were named as ‘‘situation-specific tension” and ‘‘situ-ation-specific concentration”.

Matthews, Desmond, Joyner, and Carcary (1997) revised the previous factor structure of the DBI and found five distinctdimensions of stress vulnerability. The revised version of the DBI was named as Driver Stress Inventory (DSI) and measuredfive dimensions of stress: ‘‘dislike of driving”, ‘‘aggression”, ‘‘fatigue proneness”, ‘‘hazard monitoring”, and ‘‘thrill-seeking”.The first three dimensions of the DSI predicted different types of subjective state disturbance during driving relating to anx-iety, anger and fatigue symptoms, respectively. Hazard monitoring dimension primarily reflects a coping style that aims toprevent threat by search for danger. Thrill-seeking is defined by items that describe enjoyment of danger (Matthews, 2002).

The link between driver stress and driver performance has been investigated previously (e.g., Evans, Palsane, & Carrere,1987; Magnavita et al., 1997; Stokols, Novaco, Stokols, & Campbell, 1978). Aggression, thrill-seeking and to some extent lowhazard monitoring were found to predict self-reported accident involvement. Aggression, thrill-seeking, and low dislike ofdriving were reported to be related to offences such as speeding, and to higher self-reported violations. Higher rates ofself-reported errors were associated with high aggression, thrill-seeking, dislike and fatigue proneness, and with low hazardmonitoring (Dorn & Matthews, 1995).

Studies with the DBI have indicated that the dimensions of driver stress vulnerability generalize across different cultures(Lajunen & Summala, 1995; Matthews, Tsuda, Xin, & Ozeki, 1999). Kontogiannis (2006) showed that driver stress in the formof aggression is associated with unsafe behaviours cross culturally. Some of the previous studies provided evidence on theeffects of sex and age on stress reactions of drivers. Male drivers reported comparatively higher aggression and comparativelylower overtaking tension than female drivers (Matthews et al., 1999). Simon and Corbett (1996) reported a negative relation-ship between age and measure of general stress. Similarly, Gulian et al. (1989) found that older drivers reported less stress.

1.2. Work conditions, driver stress and risky behaviours in traffic settings

Westerman and Haigney (2000) indicated that driver stress includes several facets each of which will have different im-pact on driver behaviour. Matthews et al. (1999) found driving related demands to be related to the DBI factors and driverstress was relevant to performance at work in several respects. Their results indicated that high levels of life stress wereassociated with increased accident involvement and there might be a correlation between stress at work and driver stress.They also emphasized the role of driver stress as a factor increasing the likelihood of accident involvement and its cost forcompanies. Gulian et al. (1989) found correlations between dislike of driving and reports of work stressors, like worriesabout redundancy and retirement. Matthews et al. (1999) stated that if the job involves vehicle driving, there is a possibilityof work demands to influence the drivers’ general attitudes and reactions toward driving. Similarly, Karasek and Theorell(1990) indicated that job status and work demands may influence stress outcomes interactively in such a way that peoplemay perceive the high workloads as less aversive if they have some control over work activities. Rosenbloom and Shahar(2007) reported professional drivers as being more influenced by traffic laws as compared to non-professional drivers.Hence, they are more likely to criticize those laws, to perceive them as unjust, and perhaps more likely to consciously dis-obey them.

34 B. Öz et al. / Transportation Research Part F 13 (2010) 32–40

Studies about professional drivers’ stress reactions and risky behaviours have been mostly conducted among some spe-cific groups of drivers, like taxi drivers (e.g., Burns & Wilde, 1995; Machin & De Souza, 2004; Rosenbloom & Shahar, 2007).However, comparisons of different professional groups on these factors have not been done so far.

The first aim of the present study was, therefore, to investigate the differences among different professional and non-pro-fessional driver groups in terms of stress reactions, speeding, number of penalties and accident involvement. The second aimwas to study the relationships between driver stress and risky driving including accident involvement, penalties andspeeding.

2. Method

2.1. Participants

A total of 234 male professional and non-professional drivers from four different driver groups (taxi drivers, N = 69; mini-bus drivers, N = 63; heavy vehicle drivers, N = 64; and non-professional drivers, N = 38) in Turkey participated in the study.The same recruitment methods were used to recruit the participants from different driver groups in order to eliminate thepossible sampling bias. All the participants participated in the study upon their personal acceptance. The taxi and heavyvehicle drivers were recruited first by contacting their companies and then they were individually asked to participate inthe study. As the minibus drivers are not working for a specific company, they were directly asked to participate in the studylike the non-professional drivers. All the participants were assured of anonymity and confidentiality. The mean and standarddeviation (SD) values of age, annual km, period of driving license, and number of accidents of each group of drivers can befound in Table 1.

2.2. Measures

2.2.1. The Driver Stress Inventory (DSI)Forty eight-item form of the Driver Stress Inventory (the DSI – Matthews et al., 1997) was used to measure stress

reactions of the drivers. The original factor structure of the DSI was applied in the present study. The DSI includes fivedistinct dimensions of stress vulnerability. These dimensions are dislike of driving with 12 items (e.g., I feel tense ornervous when overtaking another vehicle), aggression with 12 items (e.g., I really dislike other drivers who causeme problems), fatigue proneness with 8 items (e.g., I become inattentive to road signs when I have to drive for severalhours), hazard monitoring with 8 items (e.g., I make an effort to look for potential hazards when driving), and thrill-seeking with 8 items (I get a real thrill out of driving fast). The participants were asked to evaluate each item on a 10-point scale.

2.2.2. Demographic information formSubjects were required to provide information on age, level of education, years a full driving license is held, number of

accidents, type of accidents (i.e., information was gathered on accidents within work hours, out of work hours, both passiveand active accidents within the last three years. The question to gather the related information was: ‘‘Within the last threeyears how many times did you have an accident without considering seriousness of it?), annual km, frequency and the typeof penalties (i.e., information was gathered on parking, speed violation, overtaking and any other kind of penalties that thedriver had within the last three years. The question to gather the related information was: ‘‘How many times did you get thefollowing traffic penalties within the last three years?), and speeding (i.e., information was gathered about speeding on in-city and highway roads via the question of ‘‘Under normal weather and road conditions how fast do you drive on in-city/highway roads on average?). The total number of accidents and penalties were measured via the demographic informationform, because it was not possible to conduct more detailed accident analyses by investigating involvement in different typesof accidents.

Table 1Mean and standard deviation (SD) values of age, annual mileage, period of driving license, and number of accidents for the four driver groups involved in thestudy.

N Age Annual mileage Period of driving license Number of accidents

Mean SD Mean SD Mean SD Mean SD

Taxi 69 39.09 11.57 53.920 46.666 17.56 10.37 1.09 1.72Minibus 63 41.41 9.77 55.431 33.312 18.11 8.00 1.57 3.69Heavy vehicle 64 38.56 8.07 95.481 124.555 17.63 6.913 1.37 2.13Non-prof. 38 39.71 11.89 17.247 217.798 15.66 9.533 1.21 4.84

B. Öz et al. / Transportation Research Part F 13 (2010) 32–40 35

3. Results

3.1. Analyses of covariance (ANCOVA)

Different ANCOVAs were conducted to investigate the differences among three groups of professional drivers and thenon-professional drivers in speeding, number of penalties, accident involvement, and on the five dimensions of the DSI. Be-fore conducting the ANCOVAs, the subscales for the DSI aggression, hazard monitoring, fatigue proneness, dislike of driving,and thrill-seeking dimensions were computed. The effects of age and annual km driven were controlled in the analyses.

Results of the ANCOVAs revealed group differences in speeding, number of penalties, and stress reactions of the drivers asmeasured with the DSI dimensions. As Table 2 shows, ANCOVA results for speeding and number of penalties revealed thatthe non-professional drivers drove faster than the taxi, minibus and heavy vehicle drivers on highways, and faster than theheavy vehicle and minibus drivers on in-city roads. Besides, the minibus drivers reported higher number of penalties thanthe heavy vehicle drivers.

ANCOVA results for stress reactions indicated differences between different driver groups in three dimensions of the DSI(see Table 3). Minibus drivers were more aggressive compared to the non-professional drivers. Non-professional driverswere better in hazard monitoring in traffic compared to the minibus and heavy vehicle drivers. Finally, heavy vehicle driversreported more fatigue proneness compared to non-professional drivers. There were no significant differences among differ-ent driver groups on dislike of driving, thrill-seeking, and in accident involvement.

3.2. Bivariate correlation analyses

To investigate the relationships among the variables of interest (i.e., age, annual km, speeding on highways, speeding onin-city roads, accident involvement, number of penalties, and the DSI dimensions), different correlation analyses were con-ducted for each driver group separately. Results of the correlation analyses revealed significant differences for different dri-ver groups in terms of the relationships among the DSI dimensions, speeding on in-city roads and highways, accidentinvolvement, and number of penalties. As shown in Table 4, there was a positive relationship between speeding on in-cityroads and accident involvement for both the minibus and heavy vehicle drivers. A positive relationship was observed be-tween speeding on highways and accident involvement for only the heavy vehicle drivers. For the heavy vehicle drivers,thrill-seeking and aggression were positively related to accident involvement. For none of the driver groups investigatedin the present study, number of penalties was significantly related to any dimension of the DSI.

For both taxi and minibus drivers, there was a positive relationship between dislike of driving and speeding on in-cityroads. The same relationship for speeding on highways was found significant only for the taxi drivers. For the heavy vehicledrivers, fatigue proneness was negatively related to speeding on highways. There was a positive relationship between speed-ing on highways and thrill-seeking for the non-professional drivers.

Table 2ANCOVA results for speeding and number of penalties.

Non-professional Heavy Vehicle Taxi Minibus F

Speeding on highways 106.45a 89.35b 95.28b 96.01b 7.97**

Speeding on in-city roads 62.94a 53.72b 56.62ab 53.24b 3.80*

Number of penalties 1.54a 1.09a 1.58ab 2.39b 3.54*

Note: Bonferroni correction was used for pair wise comparisons. Mean values with different superscripts within rows are statistically different from eachother.* p < .05.** p < .001.

Table 3ANCOVA results for the dimensions of the DSI.

Non-professional Heavy vehicle Taxi Minibus F

Aggression 48.22a 53.93ab 55.71ab 62.49b 4.95*

Hazard monitoring 64.29a 51.36b 56.62ab 52.05b 4.40*

Fatigue proneness 41.56a 56.68b 50.69ab 51.46ab 3.90*

Dislike of driving 33.64 29.60 33.29 30.94 1.04Thrill-seeking 34.77 42.28 41.18 48.18 1.98

Note: Bonferroni correction was used for pair wise comparisons. Mean values with different superscripts within rows are statistically different from eachother.* p < .01.

Table 4Correlations among age, annual mileage, number of penalties, number of accidents, speeding and the DSI dimensions for the different professional and non-professional driver groups.

1 2 3 4 5 6 7 8 9 10

1. AgeTaxiMinibusHeavy vehicleNon-Professional

2. Annual mileageTaxi �0.12Minibus �0.08Heavy vehicle �0.02Non-professional 0.36*

3. Number of penaltiesTaxi 0.02 0.16Minibus �0.35** �0.09Heavy vehicle �0.25 �0.04Non-professional �0.05 0.20

4. Number of accidentsTaxi �0.25* 0.20 0.21Minibus �0.13 0.27* 0.10Heavy vehicle �0.11 0.42** 0.11Non-professional 0.08 0.49** 0.40*

5. Speeding on in-city roadsTaxi �0.24 0.13 �0.14 0.01Minibus �0.13 0.50** 0.05 0.35**

Heavy vehicle �0.10 0.25 0.16 0.45**

Non-professional �0.26 �0.14 �0.07 �0.14

6. Speeding on highwaysTaxi �0.16 0.21 0.05 0.02 0.53**

Minibus �0.28* 0.26* 0.12 0.04 0.36**

Heavy vehicle 0.13 0.06 �0.01 0.32* 0.32*

Non-professional �0.05 0.15 0.23 0.13 �0.04

7. AggressionTaxi �0.01 0.17 0.07 �0.09 �0.18 �0.20Minibus �0.08 �0.16 0.19 0.18 0.10 �0.11Heavy vehicle 0.07 0.16 0.03 0.31* 0.16 0.00Non-professional �0.26 �0.05 0.26 0.31 0.01 0.14

8. Dislike of drivingTaxi 0.06 �0.02 �0.07 0.07 0.30* 0.26* �0.40**

Minibus �0.02 0.11 �0.13 0.12 0.27* 0.25 �0.27*

Heavy vehicle �0.25* 0.26 0.05 0.09 0.21 0.06 0.12Non-professional 0.11 0.20 0.29 0.37* �0.12 �0.03 0.03

9. Hazard monitoringTaxi �0.16 0.14 0.01 0.22 0.20 0.07 �0.64** 0.38**

Minibus �0.08 0.31* �0.19 0.08 0.08 0.17 �0.72** 0.24Heavy vehicle �0.23 �0.10 0.22 �0.09 0.07 �0.11 �0.46** 0.21Non-professional �0.01 �0.04 0.26 0.20 �0.25 0.03 0.03 0.35*

10. Fatigue pronenessTaxi 0.43** �0.25 0.01 �0.07 �0.14 �0.05 0.26* 0.00 �0.29*

Minibus 0.23 �0.18 0.03 �0.08 0.16 �0.09 0.59** 0.06 �0.51**

Heavy vehicle 0.18 �0.25 �0.14 �0.17 �0.11 �0.30* 0.41** �0.15 �0.42**

Non-professional �0.01 0.10 �0.11 0.00 0.06 �0.25 0.19 0.33* �0.05

11. Thrill-seekingTaxi �0.11 0.04 0.10 0.06 �0.17 �0.10 0.78** 0.54** �0.64** 0.17Minibus �0.05 �0.27* 0.23 0.03 �0.04 �0.14 0.87** �0.39** �0.74** 0.47**

Heavy vehicle 0.12 0.19 �0.12 0.31* �0.01 �0.20 0.70** �0.03 �0.52** 0.52**

Non-professional �0.37* �0.02 0.04 0.29 �0.23 0.37* 0.42** �0.01 0.14 0.13

* p < .05.** p < .01.

36 B. Öz et al. / Transportation Research Part F 13 (2010) 32–40

The mentioned significant correlations obtained for drivers of different driver groups were compared to test their differ-ences from each other by using the formula of Cohen, Cohen, West, and Aiken (2003). Results showed that none of the cor-relations were significantly different from each other.

As Table 4 shows, bivariate correlation analyses also indicated significant relationships among different DSI dimensionsfor different driver groups. For example, aggression and thrill-seeking were positively related to each other for all driver

B. Öz et al. / Transportation Research Part F 13 (2010) 32–40 37

groups. For taxi and minibus drivers dislike of driving and aggression were negatively related to each other. For taxi, minibus,and heavy vehicle drivers, aggression and fatigue proneness were positively related to each other, whereas aggression andhazard monitoring were negatively related to each other for the same driver groups. Comparisons of correlations indicatedsignificant differences for the relationship between aggression and thrill-seeking dimensions of the DSI between taxi andnon-professional drivers (z = 2.84, p < 0.001), minibus and heavy vehicle drivers (z = 2.59, p < 0.01), minibus and non-profes-sional drivers (z = 4.21, p < 0.001), and heavy vehicle and non-professional drivers (z = 1.99, p < 0.01). The correlations be-tween aggression and fatigue proneness dimensions of the DSI were significant from each other for the taxi and minibusdrivers (z = 2.29, p < 0.01). Lastly, a significant difference was observed in the correlations between aggression and hazardmonitoring dimensions of the DSI for the minibus and heavy vehicle drivers (z = 2.08, p < 0.05). These results indicated thatthese differences between correlations in different groups are not by chance.

3.3. Regression analyses

To test the relationship among the DSI dimensions and variables of interest (i.e., number of accidents, number of penaltiesand speeding), two sets of regression analyses were conducted for different driver groups. To predict the number of accidentsand the number of penalties, logistic regression was used by recoding ‘‘number of accidents” and ‘‘number of penalties” vari-ables in binary variables (the subjects with no accidents/penalties were recoded as ‘‘0”, and the subjects with one or moreaccidents/penalties were recoded as ‘‘1”). Hierarchical regression analyses were used for predicting speeding on in-city roadsand on highways. During the regression analyses, originally categorical driver group variable (i.e., being a taxi, minibus, hea-vy vehicle or non-professional driver) was recoded in a dummy variable by making the non-professional driver group as thereference category.

In the first set of regression analyses, the number of accidents and the number of penalties were used as dependentvariables. In these logistic regression analyses, age and annual km were entered in the first step, and the driver groupwas entered in the second step. After controlling the effects of these variables the effects of the DSI dimensions were enteredin the third step. As Table 5 shows, aggression, dislike of driving, and hazard monitoring dimensions of the DSI were relatedto accident involvement. Accidents were more frequently reported by the drivers with high aggression, dislike of driving andhazard monitoring scores. None of the DSI dimensions were related to the number of penalties.

In the second set of regression analyses, speeding on in-city roads and on highways were used as dependent variables. Inthese hierarchical regressions, age and annual km were entered in the first step and driver group was entered in the second

Table 5Regression analyses on the number of accidents and number of penalties with the DSI dimensions.

Step Independent variables B SE Wald test (z-ratio) EX(B)

Number of accident as the dependent variable1 Age �0.01 0.02 0.11 0.99

Annual km driven 0.00 0.00 6.91** 1.00

2 Taxi �0.06 0.51 0.01 0.95Minibus 0.27 0.51 0.27 1.31Heavy vehicle �0.13 0.56 0.05 0.88(Constant) �5.92 1.72 11.90 0.00

3 Aggression 0.04 0.02 6.04** 1.04Hazard monitoring 0.04 0.01 9.10*** 1.04Fatigue proneness �0.01 0.01 0.99Dislike of driving 0.03 0.02 1.09 1.03Thrill-seeking 0.01 0.01 4.34* 1.01

0.98

Number of penalties as the dependent variable1 Age �0.03 0.02 3.40 0.97

Annual km driven 0.00 0.00 0.32 1.00

2 Taxi 0.05 0.50 0.01 1.05Minibus 0.72 0.54 1.80 2.06Heavy vehicle �0.77 0.53 2.08 0.46(Constant) �0.39 1.56 0.06 0.68

3 Aggression 0.20 0.02 1.59 1.02Hazard monitoring 0.02 0.01 1.90 1.02Fatigue proneness �0.01 0.01 0.26 0.99Dislike of driving 0.01 0.01 0.73 1.01Thrill-seeking 0.01 0.01 0.36 1.01

Note: df value is 1 and it is the same for all models.* p < .05.** p < .01.*** p < .001.

Table 6Hierarchical regression analyses on speeding on in-city roads and on highways with the DSI dimensions.

Step Independent variables R2 Adj. R2 R2 change F df D

Speeding on in-city roads as a dependent variable1 Age 0.07 0.06 0.07 7.09 2 �0.25***

Annual km driven 0.04

2 Taxi �0.20*

Minibus 0.12 0.10 0.05 5.24 3 �0.30**

Heavy vehicle �0.29**

3 Aggression 0.25*

Hazard monitoring �0.03Fatigue proneness 0.19 0.15 0.07 4.47 5 �0.00Dislike of driving 0.14Thrill-seeking �0.33**

Speeding on highways as a dependent variable1 Age 0.07 0.06 0.07 7.88 2 �0.27***

Annual km driven 0.03

2 Taxi �0.28**

Minibus 0.16 0.14 0.09 7.62 3 �0.24**

Heavy vehicle �0.43***

3 Aggression �0.14Hazard monitoring 0.08Fatigue proneness 0.22 0.18 0.06 5.33 5 �0.13Dislike of driving 0.07Thrill-seeking 0.08

* p < .05.** p < .01.*** p < .001.

38 B. Öz et al. / Transportation Research Part F 13 (2010) 32–40

step to the analysis. After controlling the effects of these variables, the DSI dimensions were forced into the model. As Table 2shows, some high correlations were observed between some dimensions of the DSI (e.g., between thrill-seeking and aggres-sion dimensions). For this reason, diagnostic statistics for multicollinearity were also calculated but no signs of multicollin-earity were found (for criteria, see Tabachnick & Fidell, 2007). Hence, all the variables were included in the hierarchicalregression analyses.

The results of the hierarchical regression analyses indicated that aggression and thrill-seeking were related to speedingon in-city roads. Drivers reported higher speeds on in-city roads when they have high aggression scores whereas slowerspeeds on in-city roads were reported by the drivers with higher thrill-seeking scores. None of the DSI dimensions were re-lated to speeding on highways (see Table 6 for hierarchical regression results).

4. Discussion

In the present study, differences among different professional and non-professional driver groups on stress reactions,speeding, number of penalties and accident involvement were investigated. Moreover, the present study investigated differ-ent professional and non-professional driver groups in terms of their relation with the DSI dimensions, speeding, number ofpenalties and accident involvement. Speeding is especially chosen as one of the variables of interest because the previousliterature indicated it as a major factor in road safety and it has direct and causal relationship with accident involvement(e.g., Aartsand & van Schagen, 2006; Carstenand & Tate, 2005).

ANCOVA results indicated that differences between different driver groups on the DSI dimensions showed different pat-terns. The results supported earlier findings indicating that professional and non-professional driver groups differ in theirstress reactions and risky driving behaviours (e.g., Matthews et al., 1999; Rosenbloom & Shahar, 2007). In addition, differ-ences between different professional driver groups on these factors were found. The present study indicated that thenon-professional drivers drove faster on both in-city roads and on highways as compared to the taxi, minibus and the heavyvehicle drivers. These findings make sense because the nature of the job for minibus, heavy vehicle and taxi drivers and thetypes of vehicles they drive might have a strong effect on speeds of professional drivers in addition to the effects of trafficrules/regulations, and the roles of company/organization that employ them (e.g., Caird & Kline, 2004). In addition to trafficconditions, taxi, minibus and heavy vehicle drivers use lower driving speed as the nature of their job requires them to stay ata constant low speed level, and to stop very frequently while driving.

The minibus drivers reported more penalties compared to the heavy vehicle drivers who participated in the study. Thisfinding also seems meaningful, as minibus drivers’ job is different from that of heavy vehicle drivers at some points. Minibusdrivers are transporting passengers; in some places they carry about fifteen people at once and while driving they have tostop many times for passengers to leave or get in. There are also some extra traffic rules for minibus drivers in Turkey, they

B. Öz et al. / Transportation Research Part F 13 (2010) 32–40 39

have different time schedules, and any problems with not obeying such rules or trying to catch up with the time schedulemight result in additional risky traffic behaviours (e.g., Karasek & Theorell, 1990). These characteristics of minibus drivers’job might cause these drivers differ from heavy vehicle drivers who do very different type of transportation and drive mostlyon highways, not in-city roads.

Another important issue is the difference between professional and non-professional drivers on the DSI dimensions. AN-COVAs conducted in the present study showed no differences in aggressive behaviour between different professional drivergroups. However, ANCOVAs indicated minibus drivers as being more aggressive in traffic as compared to the non-profes-sional drivers. Since their job is driving, professional drivers are more prone to experience difficulties and stress causedby traffic, as a result the probability that they will reflect this stress in terms of aggression might be higher (Caird & Kline,2004; Westerman & Haigney, 2000). It is also not surprising that non-professional drivers are better in hazard monitoring ascompared to the minibus and heavy vehicle drivers. Among those involved in the present study, minibus drivers and heavyvehicle drivers had the highest exposure to great variety of traffic conditions because of the nature of their job. Thus, it isreasonable to think that professional drivers get used to risks in traffic and perceive certain traffic situations less risky be-cause of extensive exposure: professional drivers simply get ‘‘desensitized” to traffic hazards because of continuous expo-sure. As their hazard monitoring increases, the professional drivers’ (taxi and minibus drivers in this case) frequency ofspeeding on highway increases. Although the correlation results do not allow conclusions about the directions of the asso-ciations, having this significant correlation for just these two groups of drivers might indicate that being exposed to the dan-gerous stimuli continuously might cause desensitization toward it.

The difference between non-professional drivers and the professional drivers in terms of fatigue proneness they have re-ported is another important finding of the present study. Taxi, minibus and heavy vehicle drivers reported significantly morefatigue as compared to the non-professional drivers. This finding supports the results of the previous studies. For example,Feyer and Williamson’s (1995) results showed that truck drivers were more prone to experience fatigue during long tripseven when compared to coach drivers. Professional drivers are more exposed to traffic for long hours what might make thosedrivers more prone to fatigue.

Results of the regression analyses revealed that aggressive drivers have higher speed on in-city roads and they are in-volved in higher number of accidents. A similar result was found when the correlations for different driver groups wereinvestigated: aggression and number of accidents were positively related among heavy vehicle drivers. Previous studies havereported an association between aggression and some deliberate violations like speeding (e.g., Matthews et al., 1997). Sim-ilarly, the regression analyses revealed that drivers with high dislike of driving scores had been involved in higher number ofaccidents. According to Matthews (2001), high dislike of driving scores may interfere with task performance because it mightgenerate negative mood states. Correlation results of the present study also revealed a positive relationship between dislikeof driving and speeding for both taxi and minibus drivers.

Two regression analyses results were unexpected. By nature, high scores on the thrill-seeking and hazard monitoringdimensions of the DSI should be related to more risky driving style and being aware of danger faster, respectively. However,the results of the present study indicated that drivers with high thrill-seeking scores drove slower on in-city roads and thatthe drivers with high hazard monitoring scores were involved in higher number of accidents. There might be different rea-sons for these unexpected results. For example, effects of some other factors, like risk perception, overestimation of hazardperception, and the mixed nature of the sample including very different types of drivers might explain these results. Highlevel of thrill-seeking could be expressed as high speeding only on the certain types of roads (i.e., motorway).

One of the main limitations of the current study is the cross-sectional design. For example, there might be confoundingfactors (e.g., environmental factors) that only affect one of the groups and, thus, lead to certain biases. Another methodolog-ical concern is related to causality: cross-sectional studies can not mostly determine causality. In self-report data collection,the participants might not fully and/or correctly remember the information they are supposed to remember. Thus, theymight give misleading or socially desirable answers.

In conclusion, the present study showed that professional drivers are more prone to stress reactions in traffic and to com-mit risky traffic behaviours. These findings indicate that different types of countermeasures should be applied among pro-fessional drivers than among general driver population. Professional drivers work for organizations with specific rules andregulations. These rules and regulations should aim at reducing the stress level of professional drivers to a non-harmful level.For example, work hours could be arranged so that drivers can drive without feeling too tired and tense. These arrangementscan be made for drivers (e.g., heavy vehicle drivers) working for private or governmental companies with well-establishedorganizational cultures. Such practices and regulations, however, might not work in companies with a poor organizationalculture characterized by solving the problems ‘‘on-the-spot” without commitment to safety rules and practices. Future stud-ies investigating driver behaviours of professional drivers should take into account the possible effects of some organiza-tional factors (e.g., such as the safety culture of the organization) besides other factors as suggested in some previousstudies (e.g., Varonen & Mattila, 2000; Vredenburgh, 2002).

Acknowledgements

This work has been supported by EU Marie Curie Transfer of Knowledge program (‘‘SAFEAST” Project No: MTKD-CT-2004-509813).

40 B. Öz et al. / Transportation Research Part F 13 (2010) 32–40

References

Aartsand, L., & van Schagen, I. (2006). Driving speed and the risk of road crashes: A review. Accident Analysis and Prevention, 38(2), 215–224.Burns, P. C., & Wilde, G. J. S. (1995). Risk taking in male taxi drivers: Relationships among personality, observational data and driver records. Personality and

Individual Differences, 18, 267–278.Caird, K. J., & Kline, T. J. (2004). The relationships between organizational and individual variables to on-the-job driver accidents and accident-free

kilometers. Ergonomics, 47(15), 1598–1613.Carstenand, O. M. J., & Tate, F. N. (2005). Intelligent speed adaptation: Accident savings and cost–benefit analysis. Accident Analysis and Prevention, 37(3),

407–416.Charbotel, B., Chiron, M., Martin, J. L., & Bergeret, A. (2001). Work-related road crashes in France. European Journal of Epidemiology, 17(8), 773–778.Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Erlbaum.Dorn, L., & Brown, B. (2003). Making sense of invulnerability in a qualitative study of police drivers. Safety Science, 41(10), 837–859.Dorn, L., & Matthews, G. (1995). Prediction of mood and risk appraisals from trait measures: Two studies of simulated driving. European Journal of

Personality, 9, 25–42.Evans, G. E., Palsane, M. N., & Carrere, S. (1987). Type A behaviour and occupational stress: A cross-cultural study of blue-collar workers. Journal of

Personality and Social Psychology, 52, 1002–1007.Feyer, A. M., & Williamson, A. M. (1995). The influence of operational conditions on driver fatigue in the long distance road transport industry in Australia.

International Journal of Industrial Ergonomics, 15(4), 229–235.Glendon, A. I., Dorn, L., Matthews, G., Gulian, E., Davies, D. R., & Debney, L. M. (1993). Reliability of the driver behaviour inventory. Ergonomics, 36, 719–726.Gulian, E., Matthews, G., Glendon, A. I., Davies, D. R., & Debney, L. M. (1989). Dimensions of driver stress. Ergonomics, 32, 585–602.Karasek, R., & Theorell, T. (1990). Healthy work: Stress, productivity and reconstruction of working life. New York: Wiley.Kontogiannis, T. (2006). Patterns of driver stress and coping strategies in a Greek sample and their relationship to aberrant behaviours and traffic accidents.

Accident Analyses and Prevention, 38, 913–924.Lajunen, T., & Summala, H. (1995). Driving experience, personality and skill and safety motive dimensions in drivers’ self-assessments. Personality and

Individual Differences, 19, 307–318.Machin, M. A., & De Souza, J. M. D. (2004). Predicting health outcomes and safety behaviour in taxi drivers. Transportation Research Part F, 7, 257–270.Magnavita, N., Narda, R., Sani, L., Cabone, A., De Lorenzo, G., & Sacco, A. (1997). Type A behaviour pattern and traffic accidents. British Journal of Medical

Psychology, 70, 103–107.Matthews, G. (2001). A transactional model of driver stress. In P. A. Hancock & P. A. Desmond (Eds.), Stress, workload and fatigue (pp. 133–166). New Jersey:

Lawrence Erlbaum Association.Matthews, G. (2002). Towards a transactional ergonomics for driver stress and fatigue. Theoretical Issues in Ergonomics Science, 3(2), 195–211.Matthews, G., Desmond, P. A., Joyner, L. A., & Carcary, B. (1997). A comprehensive questionnaire measure of driver stress and affect. In J. A. Rothengatter & E.

C. Vaya (Eds.), Traffic and transport psychology: Theory and application (pp. 317–324). Amsterdam: Elsevier.Matthews, G., Dorn, L., & Glendon, A. I. (1991). Personality correlates of driver stress. Personality and Individual Differences, 12, 535–549.Matthews, G., Tsuda, A., Xin, G., & Ozeki, Y. (1999). Individual differences in driver stress vulnerability in a Japanese sample. Ergonomics, 42(3), 401–415.Matthews, G., & Wells, A. (1996). Attentional processes, coping strategies and clinical intervention. In M. Zeindler & N. S. Endler (Eds.), Handbook of coping:

Theory, research, application (pp. 573–601). Wiley: New York.Rosenbloom, T., & Shahar, A. (2007). Differences between taxi and nonprofessional male drivers in attitudes toward traffic violation penalties. Transportation

Research Part F, 10, 428–435.Salminen, S., & Lähdeniemi, E. (2002). Risk factors in work-related traffic. Transportation Research Part F, 5, 77–86.Simon, F., & Corbett, C. (1996). Road traffic offending, stress, age, and accident history among male and female drivers. Ergonomics, 39, 757–780.Stokols, D., Novaco, R. W., Stokols, J., & Campbell, J. (1978). Traffic congestion, Type A behaviour, and stress. Journal of Applied Psychology, 63, 467–480.Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Pearson/Allyn & Bacon.Toscano, G., & Windau, J. (1994). The changing character of fatal work injuries. Monthly Labour Review, 17, 28.Varonen, U., & Mattila, M. (2000). The safety climate and its relationship to safety practices, safety of the work environment and occupational accidents in

eight wood-processing companies. Accident Analysis and Prevention, 32, 761–769.Vredenburgh, A. G. (2002). Organizational safety: Which management practices are most effective in reducing employee injury rates? Journal of Safety

Research, 33, 259–276.Westerman, S. J., & Haigney, D. (2000). Individual differences in driver stress error and violation. Personality and Individual Differences, 29, 981–998.