Accident Analysis and Prevention 36 (2004) 323332 [612753]

Accident Analysis and Prevention 36 (2004) 323–332
The multidimensional driving style inventory—scale
construct and validation
Orit Taubman-Ben-Aria,∗, Mario Mikulincerb, Omri Gillathb
aSchool of Social Work, Bar-Ilan University, Ramat Gan 52900, Israel
bDepartment of Psychology, Bar-Ilan University, Ramat Gan 52900, Israel
Received 2 June 2002; received in revised form 2 December 2002; accepted 11 December 2002
Abstract
Two studies were conducted in order to develop a multidimensional instrument of driving style. In Study 1, we developed a self-report
scaleassessingfourbroaddomainsofdrivingstyle—themultidimensionaldrivingstyleinventory(MDSI).Afactoranalysisrevealedeightmain factors, each one representing a specific driving style—dissociative, anxious, risky, angry, high-velocity, distress reduction, patient,and careful. In addition, significant associations were found between the eight factors, on the one hand, and gender, age, driving history,and personality measures of self-esteem, need for control, impulsive sensation seeking, and extraversion, on the other. In Study 2, furtherassociations were found between the eight driving style factors and measures of trait anxiety and neuroticism. The discussion focused onthe validity and utility of a multidimensional conceptualization of driving style.© 2003 Elsevier Ltd. All rights reserved.
Keywords: Driving style; Personality traits; Reckless driving
1. Introduction
Inthelastseveralyears,therehasbeenagrowingconcern
about the harsh consequences of driving and an increasedlevelofinterestinthetrafficsafetyproblemofcaraccidents(Harré, 2000; West et al., 1993 ). This line of research has
mainlyfocusedonhumanfactorsthatareinvolvedincarac-cidents, such as sociodemographic and general personalityfactorsaswellasdriving-specificskills,attitudes,andbehav-iors (e.g. Beirness, 1993; Garrity and Demick, 2001; Jonah,
1997; West et al., 1993 ). The current study follows this line
of research and mainly focuses on the conceptualization ofa person’s habitual driving style as a driving-specific factorthat can directly explain involvement in car accidents andmediate the effects of more general sociodemographic andpersonality factors.
A review of the literature indicates that previous research
has mostly dealt with the association between various so-ciodemographic factors (e.g. age, gender, experience) orgeneral personality traits (e.g. sensation seeking, type A/Bpersonality, locus of control) and involvement in car acci-dents(e.g. FurnhamandSaipe,1993 ).Inthiscontext,gender
andagehaveconsistentlybeenfoundtoberelatedtodriver’s
∗Corresponding author. Tel.: +972-3-531-8066; fax: +972-3-534-7228.
E-mail address: [anonimizat] (O. Taubman-Ben-Ari).accident risk and traffic violations (e.g. Lawton et al., 1997;
Maycock et al., 1991; Westerman and Haigney, 2000 ). Al-
mosteverymeasureofinvolvementinfatalcrashesrecordedin the USA during the 1980s showed rates for men approx-imately double those for women ( Evans, 1991 ), as well as
increased crash involvement and a higher rate of risk takingwhile driving among younger drivers (e.g. Glendon et al.,
1996; Matthews and Moran, 1986; Maycock et al., 1991 ).
Personality traits have been also shown to be related to
risky driving and crash involvement. In this context, traitsof sensation seeking, impulsiveness, and thrill and adven-ture seeking seem to be the strongest predictors of recklessdriving and involvement in car accidents. Specifically, thesetraits have been consistently associated with engagementin risky driving practices, such as speeding or impaireddriving, and involvement in traffic violations and accidents(e.g.Arnett et al., 1997; Beirness and Simpson, 1988, 1990;
Donovan et al., 1990; Jonah, 1997; Trimpop and Kirkcaldy,1997;ZuckermanandNeeb,1980 ).Accordingly,somestud-
ies have reported that the trait of desire for control is alsorelated to reckless driving and car accidents (e.g. Horswill
and McKenna, 1999; Trimpop and Kirkcaldy, 1997 ). With
regardtotraitsofextraversionandneuroticism( Eysenckand
Eysenck,1975 ),thefindingsarelessconclusive.Ontheone
hand, some studies have found significant associations be-tweenthesetraitsandbothnumberofcrashesandviolations
0001-4575/$ – see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0001-4575(03)00010-1

324 O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332
(e.g.Fine, 1963; Renner and Anderle, 1999; Shaw and
Sichel, 1971 ). On the other hand, these findings were sig-
nificant only for men and additional studies have failed tofind such a relationship even among men (e.g. Matthews
et al., 1991; Wilson and Greensmith, 1983 ).
To date, most of the researchers agree that the above
reviewed findings are highly important for understandinginvolvement in car accidents, but they do not provide in-formation about the specific driving-related factors that di-rectlyunderlierecklessdriving.Inthiscontext, Elanderetal.
(1993)have argued that accident liability is related to driv-
ing skill and to driving style. By “skill” they referred to theabilities of drivers to maintain control of the vehicle andrespond adaptively to complex traffic situations. In otherwords, they refer to the driver’s performance. Driving skillis expected to improve with practice or training. By “style”they referred to the ways drivers choose to drive or habitu-ally drive. This includes choice of driving speed, headway,andhabituallevelofgeneralattentivenessandassertiveness.Driving style is expected to be influenced by attitudes andbeliefs regarding driving as well as more general needs andvalues. It is this aspect of driving that stands in the focus ofthe present investigation.
Despite the agreement about the importance of driving
style, there is no agreement about its conceptualization andmeasurement. In fact, several self-report measures of driverbehaviorandcognitiontappingverydifferentaspectsofdriv-ing have been constructed in the last years ( Westerman and
Haigney, 2000 )—Driving Behavior Inventory (DBI, Gulian
et al., 1988; Gulian et al., 1989 ), Driving Style Question-
naire (DSQ, French et al., 1993 ), The Attitudes to Driving
Violations (ADVS, West and Hall, 1997 ), Driver Behavior
Questionnaire (DBQ, Reason et al., 1990 ), Drivers Behav-
ior Questionnaire ( Furnham and Saipe, 1993 ), and Driving
Vengeance Questionnaire (DVQ, Wiesenthal et al., 2000 ).
The DSQ ( French et al., 1993 ), for example, examines
behaviors that had been shown to be related to accidentinvolvement or risky driving behavior, such as, speed, head-way (distance to the car in front), seat belt use, gap accep-tance (size of gap in the flow of traffic before attempting topull out), and traffic light violations, as well as cognitionsand attitudes that are supposed to be directly related todriving decision-making, such as feelings of control, routeplanning, and risk taking on the road. Another assessmentprocedure, the DBI ( Gulian et al., 1988, 1989 ), focuses on
driving stress and taps dimensions of driving aggression,driving alertness, dislike of driving, general driver stress,irritationwhenovertaken,andfrustrationinovertaking.Yet,the DBQ ( Reason et al., 1990 ) examines errors made while
driving,deliberateviolationsofnormalsafedrivingpractice,and harmless mistakes that result from inattention (lapses).
We believe that this diversity of conceptualizations and
measurement scales tapping driving style reflects the highlycomplex and multidimensional nature of the phenomenon.However,wealsothinkthatthestatusoftheoryandresearchin driving style enables the conceptual and empirical inte-gration of the various definitions and scales into a single,
multidimensionalconceptualizationofdrivingstyle.Onthisbasis, we reviewed the diverse scales of driving styles andconceptually analyzed the underlying factor structures ofthese scales. Even though most researchers were interestedin driving behaviors which are related to car accidents, webroadened our scope to various behaviors and habits whichare related to driving in general in order to reveal the wholerange of driving styles that can predict involvement in caraccidents.
Followingareviewoftheexistingscalesofdrivingstyles,
we hypothesize that most of the driving-specific factorsidentified in these scales can be integrated into four broadfacets: (a) reckless and careless driving style, (b) anxiousdriving style, (c) angry and hostile driving style, (d) patientand careful driving style. The reckless and careless driving
stylerefers to deliberate violations of safe driving norms,
and the seeking of sensations and thrill while driving (e.g.French et al., 1993; Reason et al., 1990 ). It characterizes
persons who drive at high speeds, race in cars, pass othercars in no-passing zones, and drive while intoxicated, prob-ably endangering themselves and others. The anxious driv-
ing stylehas commonly been examined in studies on driver
stress (e.g. Gulian et al., 1989 ) and reflects feelings of alert-
ness and tension as well as ineffective engagement in relax-ing activities during driving. The angry and hostile driving
stylerefers to expressions of irritation, rage, and hostile
attitudes and acts while driving, and reflects a tendency toact aggressively on the road, curse, blow horn, or “flash” toother drivers (e.g. Arnett et al., 1997; Donovan et al., 1988 ).
Thepatientandcarefuldrivingstyle reflectsawell-adjusted
driving style that has received less attention in previousstudies (e.g. French et al., 1993; Harré, 2000 ). This style
refers to planning ahead; attention, patience, politeness, andcalmness while driving; and keeping the traffic rules.
After conceptualizing the above four domains of driving
styles,ournextstepsweretobuildaself-reportscaleforas-sessing these domains, to examine whether the factor struc-ture of this scale validated the hypothesized four domains,andtoexploretheassociationsbetweenthesedomains,driv-ing behaviors, and sociodemographic and personal factors.This research program can provide important informationon the usefulness of a multidimensional scale for assess-ing driving styles and clarify the associations of these styleswith a host of other variables.
2. Study 1
The first study was intended to construct a self-report
instrument in order to assess the four domains of drivingstyle and their relevance for examining variations in his-tory of driving in general and reckless driving in particular.We drew on existing theoretical and empirical literatureto identify four domains of driving style. Next, we con-structed a measure to assess driving styles in these domains

O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332 325
by adapting items from several existing measures, such
as the DSQ ( French et al., 1993 ), DBQ ( Reason et al.,
1990), DBI (Gulian et al., 1988, 1989 ), and by writing ad-
ditional original items. Then, we examined the associationsbetween these styles and measures of reckless driving, so-ciodemographic factors (gender, age, level of education),and personality traits (self-esteem, need for control, impul-sive sensation seeking, and extraversion). Since previousstudies have generally failed to take into account annualmileage, which has been found related to accident rates andto propensity to drive fast (e.g. French et al., 1993; Quimby
et al., 1986 ), we controlled for variations in annual mileage
while examining the above associations.
2.1. Method
2.1.1. Participants
Three hundred and twenty eight participants from var-
ious geographical areas in Israel who owned a drivinglicense and drove on a regular basis volunteered to partic-ipate in the study. These participants were sampled via the“snowball” technique: the questionnaires were given to aninitial sample of university and college students, who askedfriends, acquaintances, and family members to completethe questionnaire. The sample consisted of 220 women and108 men, ranging in age from 19 to 70 (mean =31.78,
S.D.=13.31). Sixty-two percent of them ( N=189) were
university students, 12% completed elementary school, and26% completed high school education.
2.1.2. Procedure and measures
Participants were asked to complete a packet of ques-
tionnaires. The questionnaires were presented in a randomorder across participants. The packet included scales tap-ping driving style, self-esteem, desire for control, impulsivesensation seeking, extraversion, and driving behaviors.
Driving style was assessed by the multidimensional
driving style inventory (MDSI) , which has been especially
constructed for this study in order to tap the four hypoth-esized domains of driving styles. Participants were askedto read each item and to rate the extent to which it fitstheir feelings, thoughts, and behavior during driving on a6-point scale, ranging from “ not at all” (1) to “ very much ”
(6).
1Originally, 20 items were written to assess each of the
four domains. This 80-item version was administrated to apilot sample of 500 participants (354 women and 146 men,ranging in age from 19 to 42, median =28), most of them
university students. Following item and exploratory factoranalyses, we retain 44 items that have an adequate normaldistributionandgoodpsychometricfeatures.These44itemsbecame the final version of the MDSI and all the statisticalanalyses were conducted on this version of the scale.
1The full questionnaire can be found on the web at: http://members.
tripod.com/drive10 /.Global self-esteem was assessed by Rosenberg’s (1979)
10-item scale. Participants rated their agreement with eachitem on a 4-point scale, ranging from 1 ( strongly disagree)
t o4(strongly agree ). In the current sample, the Cronbach’s
αfor the 10 items was high (0.86). Then, we averaged
the 10 items, with higher scores indicating more positiveself-esteem. Desire for control was assessed by Burger and
Cooper’s (1979) 20-item scale, which taps need for control
in daily activities. Participants are required to respond on a7-point Likert type scale, ranging from 1 ( never)t o7(al-
ways). The Cronbach’s αfor the 20 items was acceptable
(0.77). Thus, we averaged all items into a single score, withhigher scores representing higher desire for control.
Impulsive sensation seeking was assessed by Zuckerman
et al. (1993) 19-item scale, which taps needs for stimula-
tion and sensation as well as impulsiveness and risk takingin decision-making. The Cronbach’s αfor the 19 items was
acceptable (0.80), thus we averaged all items into a singlescore,withhigherscoresrepresentinghigherimpulsivesen-sation seeking. Extraversion was assessed by the Extraver-sionsubscaleoftheEysenckPersonalityInventory( Eysenck
and Eysenck, 1967 ), which was composed of 23 items that
could be answered yesorno. The Cronbach’s αfor the 23
items was acceptable (0.79). Thus, we averaged all itemsinto a single score, with higher scores representing higherextraversion.
Attheendofthesequestionnaires,participantswereasked
to provide sociodemographic information as well as infor-mation about exposure, by reporting on the average amountof kilometers driven per day during the week and duringweekends;involvementincaraccidents(thelifetimenumberof involvement in car accidents), and the lifetime frequencyof 13 driving offenses (e.g. speeding, crossing in red light).
2.2. Results and discussion
2.2.1. MDSI factors
To determine whether the 44 MDSI items fell into distin-
guishable domains, a factor analysis with Varimax rotationwas conducted on these 44 items. The factor analysis re-vealedeightmainfactors(eigenvalue >1),whichexplained
56% of the variance of the 44 items. Table 1presents load-
ings of the items in each of the factors. Factor 1 explained21%ofthevariance(Cronbach’s /H9251=0.82)andconsistedof
8 items that load high (greater than 0.40) on the factor. Allthese items tap a person’s tendency to be easily distractedduring driving, to commit driving errors due to this distrac-tion, and to display cognitive gaps and dissociations duringdriving. On this basis, we labeled this factor as “dissocia-tive driving style”. Factor 2 explained 10% of the variance(Cronbach’s α=0.82) and consisted of 7 items that load
high on the factor. All these items tap a person’s tendencyto feel distress during driving, to display signs of anxietydue to the driving situation, and to express doubts and lackof confidence about his or her driving skills. On this basis,we labeled this factor as “anxious driving style”.

326 O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332
Table 1
Factor model coefficients of the multidimensional driving style inventory
Factors and items Loading
Factor 1: dissociative driving style
[30] misjudge the speed of an oncoming vehicle when passing 0.76
[34] intend to switch on the windscreen wipers, but switch on the lights instead 0.70[27] forget that my lights are on full beam until flashed by another motorist 0.69[39] nearly hit something due to misjudging my gap in a parking lot 0.68[36] plan my route badly, so that I hit traffic that I could have avoided 0.56[35] attempt to drive away from traffic lights in third gear (or on the neutral mode in automatic cars) 0.48[15] lost in thoughts or distracted, I fail to notice someone at the pedestrian crossings 0.48[11] I daydream to pass the time while driving 0.47
Factor 2: anxious driving style
[31] feel nervous while driving 0.75
[33] feel distressed while driving 0.75
[10] driving makes me feel frustrated 0.68
[25] it worries me when driving in bad weather 0.52
[7] on a clear freeway, I usually drive at or a little below the speed limit 0.52[4] feel I have control over driving [ −] 0.49
[40] feel comfortable while driving [ −] 0.48
Factor 3: risky driving style
[44] enjoy the excitement of dangerous driving 0.83
[6] enjoy the sensation of driving on the limit 0.82
[22] like to take risks while driving 0.80
[24] like the thrill of flirting with death or disaster 0.66
[20] fix my hair/ makeup while driving 0.45
Factor 4: angry driving style
[12] swear at other drivers 0.72
[3] blow my horn or “flash” the car in front as a way of expressing frustrations 0.72[28] when someone does something on the road that annoys me, I flash them with the high beam 0.73[43] honk my horn at others 0.67
[19] when someone tries to skirt in front of me on the road, I drive in an assertive way in order to prevent it 0.48
Factor 5: high-velocity driving style
[16] in a traffic jam, I think about ways to get through the traffic faster 0.72[9] when in a traffic jam and the lane next to me starts to move, I try to move into that lane as soon as possible 0.71[17] when a traffic light turns green and the car in front of me doesn’t get going immediately, I try to urge the driver to move on 0.59[2] purposely tailgate other drivers 0.58
[32] get impatient during rush hours 0.46
[5] drive through traffic lights that have just turned red 0.40
Factor 6: distress-reduction driving style
[37] use muscle relaxation techniques while driving 0.73
[8] while driving, I try to relax myself 0.71
[1] do relaxing activities while driving 0.63
[26] mediate while driving 0.56
Factor 7: patient driving style
[18] at an intersection where I have to give right-of-way to oncoming traffic, I wait patiently for cross-traffic to pass 0.68[23] base my behavior on the motto “better safe than sorry” 0.52
[13] when a traffic light turns green and the car in front of me doesn’t get going, I just wait for a while until it moves 0.49[38] plan long journeys in advance 0.49
Factor 8: careful driving style
[42] tend to drive cautiously 0.56
[14] drive cautiously 0.55
[41] always ready to react to unexpected maneuvers by other drivers 0.51[21] distracted or preoccupied, and suddenly realize the vehicle ahead has slowed down, and have
to slam on the breaks to avoid a collision [ −]0.51
[29] get a thrill out of breaking the law [ −] 0.51
Numbers in brackets represent the order of the items in the scale.
[−] reversed item.
The detailed loadings of the 44 items on the 8 factors are available upon request from the authors.

O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332 327
Factor 3 explained 6% of the variance (Cronbach’s α=
0.83) and consisted of 5 items that load high on the fac-
tor. All these items tap a person’s seeking for stimulation,sensation, and risk during driving and his or her tendencyto take risky driving decisions and to engage in risky driv-ing. On this basis, we labeled this factor as “risky drivingstyle”. Factor 4 explained 5% of the variance (Cronbach’sα=0.80) and consisted of 5 items that load high on the
factor.Alltheitemstapaperson’stendencytobehostileto-wards other drivers as well as behave aggressively and feelintense anger while driving. On this basis, we labeled thisfactoras“angrydrivingstyle”.Factor5explained4%ofthevariance (Cronbach’s α=0.76) and consisted of 6 items
thatloadhighonthefactor.Alltheitemstapaperson’sten-dency to drive fast, to display signs of time pressure whiledriving, and to be oriented towards high velocity driving.Therefore, we labeled this factor as “high-velocity drivingstyle”.
Factor 6 explained 4% of the variance (Cronbach’s
α=0.75) and consisted of 4 items that load high on the
factor. These items tap a person’s tendency to engage inrelaxing activities during driving aimed at reducing dis-tress while driving. On this basis, we labeled this factor as“distress-reduction driving style”. Factor 7 explained 3%of the variance (Cronbach’s α=0.74) and consisted of
4 items that load high on the factor. All the items tap aperson’s tendency to be polite towards other drivers, to feelno time pressure during driving, and to display patiencewhile driving. On this basis, we labeled this factor as “pa-tient driving style”. Factor 8 explained 3% of the variance(Cronbach’s α=0.76) and consisted of 5 items that load
high on the factor. All the items tap a person’s tendencyto be careful during driving, to effectively plan his or herdriving trajectory, and to adopt a problem-solving attitudetowards driving-related problems and obstacles. On thisbasis, we labeled this factor as “careful driving style”.
As can be seen, the factor analysis revealed eight co-
herent and meaningful driving styles. Scores for each ofthe eight factors were computed by averaging items load-ing high on each factor. Pearson correlations between theeight factors revealed an interesting pattern of associations.First, significant positive associations were found betweenrisky, high-velocity, angry, and dissociative driving styles,r(s) ranging from 0.34 to 0.50, all P(s)<0.01, implying
the existence of an underlying maladaptive way of drivingthat may be theoretically associated with emotional malad-justmentaswellaswithhighlikelihoodofcaraccidentsanddriving offenses. Second, the above four maladaptive driv-ing styles were inversely and significantly associated withthe careful and patient styles, that reflect more adequate,controlled, and socially adjusted ways of driving, r(s) rang-
ing from −0.20 to −0.49, all P(s)<0.01. Third, signif-
icant positive associations were found between the carefuland patient styles, r(309)=0.21,P<0.01, as well as be-
tween the anxious and distress-reduction factors, r(309)=
0.25,P<0.01. Fourth, the anxious and dissociative styleswerealsosignificantlyassociated, r(309)=0.47,P<0.01.
Other correlations were not statistically significant.
Overall,theMDSIpresentsacomprehensive,multidimen-
sional picture of the various orientations people may adoptwhile driving. In this way, the MDSI compliments exist-ing self-report scales. Whereas these scales focus on onlyone or two of the MDSI factors (e.g. driving stress, driv-ing aggression, risky driving), the MDSI could delineate aperson’s profile across eight differentiated, and even antag-onistic, driving orientations.
2.2.2. Driving styles and sociodemographic factors
In the next step, we examined the association between
the eight driving style scores and three basic sociodemo-graphic characteristics (sex, age, education level). Genderdifferences in driving style were examined by multivariateand univariate analyses of variance (ANOVA). The mul-tivariate ANOVA revealed a significant gender difference,F(8,319)=5.39,P<0.01. Univariate ANOVAs indi-
catedthatgenderdifferencesweresignificantindissociativedriving style, F(1,326)=14.74,P<0.01, anxious driv-
ing style, F(1,326)=10.77,P<0.01, and careful driv-
ing style, F(1,326)=24.13,P<0.01. An examination
of group means (see Table 2) revealed that women scored
higher in dissociative and anxious driving styles than men.Menscoredhigherthanwomenincarefuldrivingstyle.The
Table 2
Means and S.D. of the multidimensional driving style inventory factorsaccording to gender
MDSI factors Men ( n=108) Women ( n=220)
Dissociative
Mean 1.80 2.13S.D. 0.57 0.74
Anxious
Mean 2.02 2.35S.D. 0.72 0.83
Risky
Mean 1.47 1.45S.D. 0.71 0.73
Angry
Mean 2.45 2.32S.D. 1.04 0.93
High-velocity
Mean 3.02 2.92S.D. 0.88 0.87
Distress reduction
Mean 2.31 2.48S.D. 0.94 0.82
Patient
Mean 4.74 4.72S.D. 0.97 0.68
Careful
Mean 4.60 4.19S.D. 0.65 0.68

328 O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332
same gender differences were obtained after controlling for
the amount of weekly driving.
Pearson correlations between age and the eight driving
style scores revealed the following significant associations:age was positively associated with careful and patientdriving styles, r(326)=0.17,P<0.01;r(326)=0.40,
P<0.01, and inversely associated with dissociative, angry,
anxious, risky, and high-velocity driving, r(326)=−0.39,
P<0.01;r(326)=−0.20,P<0.01,r(326)=−0.22,
P<0.01;r(326)=−0.26,P<0.01;r(326)=−0.19,
P<0.01. That is the older the participant, the higher his or
her tendency to adopt a careful and patient driving style andthe lower his or her tendency to adopt dissociative, angry,anxious, risky, or high-velocity driving styles. These corre-lations remained the same after controlling for the amountof weekly driving.
Partial correlations between education level and the eight
drivingstylescores(controllingforage)revealedsignificantassociations between education level and the endorsementof anxious and distress-reduction driving styles, r(326)=
0.18,P<0.01;r(326)=0.18,P<0.01. That is the
higher the education level of a participant, the higher his orher tendency to feel anxiety during driving and to adopt adistress-reduction style.
Overall, the findings strengthen the confidence in the
construct validity of the DSI factors. Our findings were inaccordance with the literature, which reveals that womentend to exhibit more driving stress than men and that mal-adaptive driving seems to diminish with age. These twotendencies were clearly identified by the DSI factors. First,women’s driving stress was manifested in their relativelyhigh scores in anxious and dissociative driving styles. Sec-ond, the tendency of older people to adopt more adaptiveways of driving was manifested in their relatively highscores in careful and patient driving styles as well as intheir relatively low scores in angry, anxious, dissociative,risky, and high-velocity driving styles.
2.2.3. Driving styles and personality traits
A canonical correlation between the set of the eight driv-
ing style scores and the set of the four assessed personalitytraits revealed a significant association, F(32,717)=4.64,
P<0.01 and explained 38% of the variance. Pearson cor-
relations (see Table 3) revealed the following significant
associations: first, self-esteem was significantly and pos-itively associated with careful and patient driving styles,and inversely associated with dissociative and risky drivingstyles. That is the more positive a participant’s self-esteem,the higher his or her tendency to adopt a careful and patientdriving style and the lower his or her tendency to adoptdissociative or risky driving styles. This pattern of findingsstrengthens our confidence in the validity of the MDSIas measuring adaptive and maladaptive driving styles.Self-esteem is viewed as one of the basic signs of psycho-logical adjustment and then should be positively associatedwithwell-adjustedstylesofdrivingandinverselyassociatedTable 3
Pearson correlations between driving style inventory factors and person-ality traits
MDSI factors Self-esteem Need for
controlSensation
seekingExtraversion
Dissociative −0.38∗∗−0.04 0.10 −0.23∗∗
Anxious −0.05 −0.08 −0.11 −0.22∗
Risky −0.19∗0.09 0.40∗∗−0.02
Angry −0.10 0.22∗0.13 0.14
High-velocity −0.11 0.13 0.18∗0.01
Distress reduction 0.04 0.01 0.01 0.06Patient 0.23
∗∗−0.04 −0.09 −0.16
Careful 0.27∗∗0.17∗−0.31∗∗0.02
∗P<0.01.
∗∗P<0.001.
with maladjusted ways of driving. As can be seen, the
current findings provide strong support for this hypothesis.Whereas self-esteem was positively associated with carefuland patient driving styles, the two adaptive driving styles, itwasinverselyassociatedwithdissociativeandriskydriving,which are considered maladaptive driving styles.
Second, need for control was significantly and positively
associated with angry and careful driving styles. That is thehighertheneedforcontrol,thehigherthetendencytoadoptanangryorcarefuldrivingstyle.Thispatternoffindingsre-flects that one of the psychological sources of angry drivingstyle is a strong need for control and that the frustration ofsuchaneedduringdrivingcouldresultinanger,aggression,and hostility towards other drivers. Interestingly, need forcontrol seems also to underlie careful driving style. This isan expected finding because careful driving has a planning,problem-solving facet (see items in Table 1), implying that
the driver feels that driving is under his or her control. Onthis basis, we can conclude that desire for control may haveboth positive and negative driving consequences. Whereasit could lead to a more careful driving, its frustration couldlead to angry driving.
Third, sensation seeking was significantly and positively
associated with risky and high-velocity driving styles, andinversely associated with patient driving. As expected, aperson’s global orientation towards stimulation and riskwas directly manifested in his or her responses to the MDSIitems. The higher a sensation seeking tendency, the higherthe tendency to adopt a risky and high-velocity drivingstyle—two manifestations of the need for stimulation andsensation during driving—and the lower the tendency toadopt a patient driving style—a style that is the opposite toa sensation seeking orientation.
Finally, extraversion was significantly and inversely re-
lated to dissociative and anxious driving styles. That is thehigher the extraversion, the lower the tendency to adopt adissociative driving style or to feel anxiety during driving.This pattern of findings fits extraverted people’s tendencyto take life easily and dismiss life hardships and difficulties.This personality orientation seems to reduce worries during

O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332 329
driving and the tendency to experience cognitive gaps and
dissociative states due to these worries.
Overall, the findings reveal consistent and coherent pat-
terns of associations between global personality traits andtheeightdrivingstyles.Importantly,partialcorrelationscon-trolling for age, sex, and amount of weekly driving (in km)revealedanidenticalpatternofassociationstothatpresentedinTable 3. On this basis, we can conclude that the various
MDSI factors tap meaningful constructs that are somewhatderived from global personality orientations.
2.2.4. Driving styles and self-reported driving behaviors
Inthenextstep,weexaminedtheassociationsbetweenthe
eightdrivingstylesandthreeself-reporteddrivingbehaviors:(a) the amount of weekly driving (in km), (b) the number ofcar accidents in which a participant reported he or she hadbeen involved in, and (c) the number of driving offenses aparticipant reported he or she had committed.
Pearson correlations revealed that the amount of weekly
driving was significantly and inversely related to anxiousdriving style, r(326)=−0.26,P<0.01. That is the
stronger a participant’s anxiety while driving, the less theamount of driving he or she undertook. No other significantassociations were found. As expected, persons who tend tofeel driving stress tend to avoid driving. This behavioralmanifestation of anxious driving style seems to strengthenthe construct validity of this MDSI factor.
Partial correlations (controlling for age and weekly driv-
ing) revealed that involvement in car accidents was sig-nificantly and positively associated with angry, risky, andhigh-velocity driving styles, r(324)=0.22,P<0.01;
r(324)=0.35,P<0.01;r(324)=0.26,P<0.01, and
inversely associated with careful driving style, r(324)=
−0.23,P<0.01. That is the higher a participant’s ten-
dency to adopt angry, risky, or high-velocity driving styles,the higher the number of accidents he or she reportedbeing involved in. Accordingly, the higher a participant’stendency to adopt a careful driving style, the lower thenumber of accidents he or she reported being invol-ved in.
In order to examine the contribution of the eight MDSI
factors to car accidents involvement, beyond the varianceexplainedbysociodemographicvariables(sex,age)andper-sonality traits (extraversion, desire for control, self-esteem,sensation seeking), we performed a discriminant analysis inwhich all these 14 variables were entered into the modelto discriminate between participants who reported being in-volved at least in one car accident and participants who re-ported being involved in no car accident. The standardizedcanonical coefficients revealed that beyond the contributionof sociodemographic and personality variables, some MDSIfactors still contributed to the discriminant function (coeffi-cientshigherthan0.35).Specifically,thedissociative(0.36),risky (0.49), and high-velocity (0.62) driving styles made aunique contribution to car accidents involvement. Interest-ingly,aftercontrollingfortheMDSIfactors,onlyage(0.73)and sensation seeking (0.41) made unique contributions to
car accidents involvement.
Partial correlations (controlling for age and weekly driv-
ing) revealed that the reported number of driving offenseswas significantly and positively associated with risky andhigh-velocity driving, r(324)=0.19,P<0.01;r(324)=
0.22,P<0.01. No other significant associations were
found. That is the higher a participant’s tendency to adoptrisky or high-velocity driving styles, the higher the numberof driving offenses he or she reported they had committed.
Amultipleregressionexaminingwhetherthedrivingstyle
scoressignificantlypredictedthenumberofdrivingoffensesrevealed that the set of the eight driving style scores signif-icantly predicted driving offenses, F(8,317)=3.65,P<
0.01, and explained 12% of the variance of this variable. In
addition,anotherregressionthatenteredtheeightMDSIfac-tors, sociodemographic variables (sex, age) and personalitytraits(extraversion,desireforcontrol,self-esteem,sensationseeking)asthepredictorsrevealedthathigh-velocitydrivingstyle still made a unique significant contribution ( B=0.33,
P<0.01) beyond the variance explained by sociodemo-
graphicandpersonalityscores.Thisregressionalsorevealedthat self-esteem was the single sociodemographic and per-sonality variable that made a unique contribution after con-trolling for MDSI scores ( B=0.31,P<0.01).
These findings present evidence supporting the validity
of the MDSI factors. First, the MDSI factors significantlypredicted self-reports of involvement in car accidents andthe amount of driving offenses. Second, those styles thatwere expected to reflect maladaptive ways of driving, suchas risky and high-velocity driving, significantly contributedtotheinvolvementincaraccidentsandtothecommissionofdriving offenses. Third, these styles still contributed to caraccidents involvement and driving offenses after controllingfor sociodemographic and personality variables.
2.2.5. Conclusions
In Study 1 we constructed a reliable and valid self-report
scale tapping driving styles. Findings revealed that eight in-ternally coherent factors of driving style underlie the itemsof this scale, and that these factors were significantly asso-ciatedwithrelevantpersonalitytraitsandsociodemographiccharacteristics. More importantly, findings indicated thatthese eight factors significantly predicted self-reports ofinvolvement in car accidents and commission of drivingoffenses.
3. Study 2
The aim of Study 2 was to further examine the construct
validity of the MDSI factors by focusing on maladaptivedriving styles and their associations with negative affectiv-ity.Ifanxious,dissociative,high-velocity,andangrydrivingstyles are valid manifestations of maladaptive ways of driv-ing, significant correlations should be found with measures

330 O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332
of global emotional maladjustment, such as trait anxiety
and neuroticism.
3.1. Method
3.1.1. Participants
One hundred and fifty Israeli university and college stu-
dents who had driving license and drove on a regular basis(86 women and 64 men, ranging in age from 19 to 45 years,mean=23.50, S .D.=4.01), volunteered to participate in
the study and to complete a battery of self-report scales.
3.1.2. Measures and procedure
All participants completed the 44-item version of the
MDSI (described in Study 1). In the current sample, a con-firmatory factor analysis revealed that the 8 main factorsexplained 57.2% of the variance and replicated the factorstructure described in Study 1. The internal consistency(Cronbach’s /H9251coefficients) of each of the eight MDSI fac-
tors was acceptable (ranging from 0.72 to 0.86). On this ba-sis,wecomputedeightdrivingstylescoresbyaveragingtheitems of each MDSI factor. Trait anxiety was assessed withthe trait form of the State-Trait Anxiety scale ( Spilberger
et al., 1970 ). This scale consisted of 20 statements tapping
the cognitive, affective, and behavioral manifestations ofanxiety. Participants rated the extent to which they agreedwitheachstatementona4-pointscale,rangingfrom“totallydisagree”(1)to“totallyagree”(4).TheCronbach’s /H9251coeffi-
cient for the 20 items in the current sample was high (0.90),allowing us to compute a trait anxiety score by averagingthe 20 items. Neuroticism was assessed by a short 12-itemHebrew version of the Neuroticism subscale of the EysenckPersonality Inventory ( Eysenck and Eysenck, 1967 ). In this
version,participantswereaskedtoratetheiragreementwitheachitemona5-pointscale,rangingfrom“notatall”(1)to“very much” (5). A total neuroticism score was computedby averaging the 12 items (Cronbach’s αof 0.89).
3.2. Results and discussion
A canonical correlation between the set of the eight driv-
ing style scores and the set of the two scores of negativeaffectivity revealed a significant association, F(16,272)=
3.28,P<0.01 and explained 21% of the variance. Pearson
correlations revealed the following significant associations:traitanxietywassignificantlyandpositivelyassociatedwithanxious and dissociative driving styles, r(148)=0.36,P<
0.01;r(148)=0.28,P<0.01, and inversely associated
with careful and patient driving styles, r(148)=−0.29,
P<0.01;r(148)=−0.25,P<0.01. That is the higher
the trait anxiety, the higher the tendency to adopt anxiousor dissociative driving styles and the lower the tendency toadoptcarefulorpatientdrivingstyles.Neuroticismwasalsosignificantly and positively associated with anxious and dis-sociative driving styles, r(148)=0.34,P<0.01;r(148)=
0.29,P<0.01, and inversely associated with careful driv-ing style, r(148)=−0.28,P<0.01. That is the higher the
neuroticismscores,thehigherthetendencytoadoptanxiousor dissociative driving styles and the lower the tendency toadopt a careful driving style.
Overall, this pattern of findings strengthens our confi-
denceinthevalidityoftheMDSIasmeasuringadaptiveandmaladaptive driving styles. Both trait anxiety and neuroti-cismarebasicsignsofpsychologicalmaladjustment.There-fore, they should be positively associated with maladjustedstylesofdrivingandinverselyassociatedwithwell-adjustedways of driving. As can be seen, the current findings pro-vide strong support for this hypothesis. Whereas trait anxi-ety and neuroticism were positively associated with anxiousand dissociative driving styles—two maladaptive drivingstyles,theywereinverselyassociatedwithcarefuldriving—a well-adjusted driving style.
4. General discussion
The purpose of this research was to highlight the need
for an integrative multidimensional measure of drivingstyles and to examine the usefulness and validity of sucha measure. Taken together, the two studies provide strongevidenceforthevalueofdistinguishingamongdifferentdo-mainsofdrivingstyleaswellasfortheinternalvalidityandusefulness of the MDSI for explaining variations in adap-tive and maladaptive driving behaviors. The correlationsbetween the eight MDSI factors and the assessed person-ality traits further attest to the importance of distinguishingamong different driving styles. Specifically, risky, dissocia-tive, and high-velocity styles were most closely associatedwith a cluster of maladaptive traits and a history of recklessdriving, whereas careful and patient styles were associatedwith adaptive aspects of personality and driving behavior.
Interestingly, although we hypothesized a construct in-
cluding four central domains of driving style, a factoranalysis of the MDSI provided strong evidence for aneight factor-solution. These eight factors cover the fourexpected driving style domains, while making more finedistinctions within each of the domains. Specifically, thereckless and careless driving style was represented by therisky and high-velocity MDSI factors; the anxious driv-ing style was represented by the anxious, dissociative anddistress-reduction MDSI factors; the angry and hostiledriving style was directly represented by the angry MDSIfactor; and the patient and careful driving style was rep-resented by two conceptually related MDSI factors—thecareful and patient factors. These eight internally coherentMDSI factors are compatible to our theoretical conceptual-ization as well as to previous studies on driving style. Theyalso highlight the complexity of driving style and the needfor a dimension-specific attitude when dealing with thisphenomenon.
Several findings supported the validity of the MDSI fac-
tors. First, these factors were significantly associated with

O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332 331
self-reports of involvement in car accidents and driving of-
fenses. Second, those MDSI factors that were theoreticallyexpected to reflect maladaptive ways of driving, such asangry, risky and high-velocity driving, were significantlyassociated with self-reports of more frequent car accidentinvolvement and commission of driving offenses. Third,the MDSI factor that was theoretically expected to reflectan adaptive way of driving (careful style) significantlycontributed to less involvement in car accidents.
Findings concerning the association between MDSI fac-
tors and sociodemographic variables also strengthened ourconfidence in the construct validity of the MDSI. Our find-ings were in accordance with the literature, which revealsthat women tend to exhibit more driving stress than men(e.g.Simon and Corbet, 1996 ) and that maladaptive driv-
ing seems to diminish with age (e.g. Glendon et al., 1996;
Maycock et al., 1991 ). These two tendencies were clearly
identifiedbytheMDSIfactors.First,women’sdrivingstresswas manifested in their relatively high scores in the anx-ious and dissociative MDSI factors. Second, the tendencyof older people to adopt more adaptive ways of driving wasmanifested in their relatively high scores in careful and pa-tient MDSI factors as well as in their low scores in angry,anxious,dissociative,risky,andhigh-velocityMDSIfactors.
The observed variations in MDSI factors were also
in accordance with general personality characteristics.Self-esteem, which represents a highly adaptive and healthypersonality trait ( Rosenberg, 1979 ), was positively associ-
ated with adaptive driving styles, and inversely associatedwith maladaptive driving styles. Whereas need for controlwas positively associated with the angry and careful MDSIfactors, sensation seeking was directly manifested in theendorsement of risky and high-velocity MDSI factors, andextraversion was inversely related to the dissociative andanxious driving MDSI factors. Both trait anxiety and neu-roticism, which are basic signs of psychological maladjust-ment, were positively associated with maladaptive drivingstyles.
Some limitations of the current studies should be noted.
First, the studies relied on self-reports of a person’s owndriving behavior. Future studies should attempt to repli-cate the present findings using behavioral measures, suchas observations of real-life driving or car-simulator driv-ing, and adopting a multi-method measurement approach.Second, no systematic attempt was made to examine thevalidity and usefulness of the MDSI in specific at riskpopulations. Further examinations should focus on specificat risk groups, such as recidivist traffic offenders, youngdrivers, etc. Despite these limitations, the current researchprovides important evidence regarding the usefulness ofthe MDSI for explaining reckless driving behavior. Futurestudies should assess the associations of the MDSI factorswith other relevant individual-differences factors, such aslocus of control and hardiness, and examine the dynamicsof driving style in different situational contexts, (e.g. thepresence of peers or adults in the car).In conclusion, it has been claimed that some 90% of
road-traffic accidents are caused by driver error ( Lewin,
1982).Therealchallengeisthereforetoprovideabetterun-
derstanding of the role of human factors in the causation ofroad accidents and consequently to develop effective coun-termeasures. These countermeasures may take the form ofimproved driver training and testing, education campaignsaimed at changing driving practices, legislation to controldriver behavior, and improvements in the design of roadsystems and vehicles ( Elander et al., 1993 ). We believe
that the MDSI scores can be taken as driving-specific fac-tors within a comprehensive model of reckless driving, aswell as working guidelines for the construction of effectivecountermeasures.
Acknowledgements
ThisresearchwassupportedbytheGeneralMotorsFoun-
dation.
References
Arnett, J.J., Offer, D., Fine, M.A., 1997. Reckless driving in adolescence:
“state” and “trait” factors. Accid. Anal. Prev. 29, 57–63.
Beirness,D.J.,1993.Dowereallydriveaswelive?Theroleofpersonality
factors in road crashes. Alcohol Drugs Driving 9, 129–143.
Beirness, D.J., Simpson, H.M., 1988. Lifestyle correlates of risky driving
and accident involvement among youth. Alcohol Drugs Driving 4,193–204.
Beirness, D. J., Simpson, H.M., 1990. Lifestyle and driving behaviours of
youth. In: Benjamin, T. (Ed.), Driving in Social Context. Paradigme,Paris, France, pp. 152–162.
Burger, J.M., Cooper, H.M., 1979. The desirability of control. Motivation
Emotion 3 (4), 381–393.
Donovan, D.M., Umlauf, R.L., Salzberg, P.M., 1988. Derivation of
personality subtypes among high-risk drivers. Alcohol Drugs Driving4, 233–244.
Donovan, D.M., Umlauf, R.L., Salzberg, P.M., 1990. Bad drivers:
identification of a target group for alcohol-related prevention and earlyintervention. J. Stud. Alcohol 51, 136–141.
Elander, J., West, R., French, D., 1993. Behavioral correlates of individual
differences in road-traffic crash risk: an examination of methods andfindings. Psychol. Bull. 113, 279–294.
Evans, L., 1991. Traffic Safety and the Driver. Van Nostrand Reinhold
Co Inc., New York, US.
Eysenck, H.J., Eysenck, S.B.G., 1967. Personality Structure and
Measurment. Routledge & Kegan Paul, London.
Eysenck, H.J., Eysenck, S.B.G., 1975. Manual of the Eysenck Personality
Questionnaire. Hodder and Stoughton, London.
Fine, B.J., 1963. Introversion-extraversion and motor vehicle driver
behavior. Perceptual Motor Skills 16, 95–100.
French,D.J.,West,R.J.,Elander,J.,Wilding,J.M.,1993.Decision-making
style, driving style, and self-reported involvement in road trafficaccidents. Ergonomics 36, 627–644.
Furnham, A., Saipe, J., 1993. Personality correlates of convicted drivers.
Pers. Individual Differences 14, 329–336.
Garrity, R.D., Demick, J., 2001. Relations among personality traits, mood
states, and driving behaviours. J. Adult Dev. 8, 109–118.
Glendon, A.I., Dorn, L., Davis, D.R., Matthews, G., Taylor, R.G., 1996.
Age and gender differences in perceived accident likelihood and drivercompetences. Risk Anal. 16, 755–762.

332 O. Taubman-Ben-Ari et al./Accident Analysis and Prevention 36 (2004) 323–332
Gulian, E., Glendon, I., Matthews, G., Davies, R., Debney, L., 1988.
Exploration of driver stress using self-reported data. In: Rothengatter,T., de Bruin, R. (Eds.), Road User Behaviour: Theory and Research.Van Gorcum & Co B.V., Assen, The Netherlands, pp. 342–347.
Gulian, E., Matthews, G., Glendon, A.I., Davies, D.R., Debney, L.M.,
1989. Dimensions of driver stress. Ergonomics 32, 585–602.
Harré, N., 2000. Risk evaluation, driving, and adolescents: a typology.
Dev. Rev. 20, 206–226.
Horswill, M.S., McKenna, F.P., 1999. The effect of perceived control on
risk taking. J. Appl. Psychol. 29, 377–391.
Jonah, B.A., 1997. Sensation seeking and risky driving: a review and
synthesis of the literature. Accid. Anal. Prev. 29, 651–665.
Lawton, R., Parker, D., Manstead, A.S.R., Stradling, S.G., 1997. The
role of affect in predicting social behaviors: the case of road trafficviolations. J. Appl. Social Psychol. 27, 1258–1276.
Lewin, I., 1982. Driver training: a perceptual-motor skill approach.
Ergonomics 25, 917–924.
Matthews, G., Dorn, L., Glendon, A.I., 1991. Personality correlates of
driver stress. Pers. Individual Differences 12, 535–549.
Matthews, M.L., Moran, A.R., 1986. Age differences in male drivers’
perception of accident risk: the role of perceived driving ability. Accid.Anal. Prev. 18, 299–313.
Maycock, J., Lockwood, C.R., Lester, J.F., 1991. The Accident Liability of
Car Drivers. (Rep. No. 315). Transport and Road Research Laboratory,Crowthorne, UK.
Quimby,A.R.,Maycock,J.,Carter,I.,Dixon,R.,Wall,J.,1986.Perceptual
Abilities of Accident Involved Drivers. (Rep. No. 27). Transport andRoad Research Laboratory, Crowthorne, UK.
Reason, J.T., Manstead, A., Stradling, S., Baxter, J.S., Campbell, K., 1990.
Errors and violations on the roads: a real distinction? Ergonomics 33,1315–1332.
Renner, W., Anderle, F.G., 1999. Venturesomeness and extraversion as
correlates of juvenile drivers’ traffic violations. Accid. Anal. Prevention32, 673–678.Rosenberg, M., 1979. Conceiving the Self. Basic Books, New York.
Shaw, L., Sichel, H.S., 1971. Accident Proneness: Research in the
Occurrence, Causation, and Prevention of Road Accidents. PergamonPress, New York.
Simon, F., Corbet, C., 1996. Road traffic offending, stress, age, and
accident history among male and female drivers. Ergonomics 39, 757–780.
Spilberger, C.D., Gorsuch, R.L., Lushene, R.E., 1970. Manual for the
State-Trait Anxiety Inventory. Consulting Psychologists Press, PaloAlto, CA, USA.
Trimpop, R., Kirkcaldy, B., 1997. Personality predictors of driving
accidents. Pers. Individual Differences 23, 147–152.
West, R., French, D., Kemp, R., Elander, J., 1993. Direct observation
of driving, self reports of driver behavior, and accident involvement.Ergonomics 36, 557–567.
West, R., Hall, J., 1997. The role of personality and attitudes in traffic
accident risk. Appl. Psychol. Int. Rev. 46, 253–264.
Westerman, S.J., Haigney, D., 2000. Individual differences in driver
stress, error, and violation. Pers. Individual Differences 29, 981–998.
Wiesenthal, D.L., Hennessy, D., Gibson, M.P., 2000. The Driving
Vengeance Questionnaire (DVQ): the development of a scaleto measure deviant drivers attitudes. Violence Victims 15, 115–136.
Wilson, T., Greensmith, J., 1983. Multivariate analysis of the relationship
between drivometer variables and drivers’ accident, sex, and exposurestatus. Hum. Factors 25, 303–312.
Zuckerman, M., Kuhlman, D.M., Joireman, M., Kraft, H., 1993. Five
robust questionnaire scale factors of personality without culture. Pers.Individual Differences 12 (9), 929–941.
Zuckerman, M., Neeb, M., 1980. Demographic influences in sensation
seeking, and expression of sensation seeking in religion, smoking anddriving habits. Pers. Individual Differences 1, 197–206.