Accident Analysis and Prevention 42 (2010) 20502056

Contents lists available at ScienceDirect

Accident Analysis and Prevention

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

Examining relationships between anxiety and dangerous driving

Chris S. Dula , Cristi L. Adams, Michael T. Miesner, Robin L. Leonard
Department of Psychology, East Tennessee State University, United States

a r t i c l e

i n f o

Article history:
Received 30 August 2009
Received in revised form 1 June 2010
Accepted 22 June 2010
Keywords:
Anxiety
Aggressive Driving
Dangerous driving
DUI/DWI
Impaired/Drunk Driving
Road rage

a b s t r a c t
Driving anxiety that has developed following crashes has been studied relatively frequently, but anxiety
per se and its effects on driving has not as yet garnered much attention in the literature. The current study
included 1121 participants and found higher levels of general anxiety were related to a wide variety of
dangerous driving behaviors. While there were clear and expected sex differences on many dangerous
driving variables, there were still more such differences with regard to anxiety levels and independent
of sex, higher levels of anxiety were associated with greater levels of dangerous driving. Of particular
import, it was found that the high anxiety group had caused signicantly more crashes and engaged in
more DUI episodes than the low and/or medium anxiety groups. Taken as a whole, the results suggest
there is a tremendous need for more research in the area of anxiety and dangerous driving and that
interventions for highly anxious drivers may well be warranted.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
As dened by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR, American Psychiatric Association, 2000),
several anxiety disorders are potentially pertinent to driving
(e.g., Generalized Anxiety Disorder, Panic Disorder, Post-Traumatic
Stress Disorder). Within this class of disorders, it appears that Specic Phobias developed after motor vehicle crashes (MVCs) have
been studied most extensively (Taylor et al., 2000; Ehlers et al.,
1998; Delahanty et al., 1997). The earliest such research focused
on treatments that reduced phobic anxiety by extinguishing conditioned reactions (Mowrer, 1960; Wolpe, 1958) and exposure-based
treatments have consistently proven to be effective in this regard
(e.g., Alphers et al., 2005; Ehlers et al., 1994; Llobet, 2009; Williams
et al., 1984).
MVC-related PTSD has also received empirical attention and
Ehlers et al. (1998) found enduring physical, psychological, and/or
nancial problems were not uncommon for victims of this disorder.
For many, PTSD symptoms following a MVC include psychologically re-experiencing the crash, persistent avoidance of thoughts or
situations associated with the crash, numbing of emotional responsiveness, and increased physical arousal (Beck and Coffey, 2007).
Linnell and Easton (2004) found MVC whiplash victims were more

Corresponding author at: Department of Psychology, East Tennessee State

University, POB 70649, Johnson City, TN 37614-1702, United States.
Tel.: +1 423 439 8307; fax: +1 423 439 5695.
E-mail addresses: dulac@etsu.edu (C.S. Dula), zcla25@goldmail.etsu.edu
(C.L. Adams), mmiesner@gmail.com (M.T. Miesner), robinleonard@charter.net
(R.L. Leonard).
0001-4575/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2010.06.016

likely to develop phobic avoidance of traveling as opposed to PTSD

per se, and that this avoidance can adversely affect lifestyle and
recovery (Linnell and Easton, 2004). This is consistent with Hodges
(1971) assertion that there is an association between fear of traveling in motor vehicles and whiplash injury.
Delahanty et al. (1997) found that MVC victims who did not
cause their crash reported more long-term distress and were more
likely to be diagnosed with PTSD than MVC victims who caused
their crash. A persisting elevated perception of threat from others
may be one reason why marginally more innocent victim participants developed PTSD than crash-causing participants. Kessler et
al. (1995) estimated a lifetime PTSD prevalence of 8% in the United
States, and found MVCs accounted for 19% of traumatic causal
events. However, while there is a good deal of literature on driving
phobia and crash-related PTSD, there is little research on general
anxiety and its relationship to driving.
Anxiety is commonly conceptualized either as tting in specic
diagnostic categories or as a trait lying on a continuum. General
anxiety should be differentiated from anxiety about driving specifically, which is not uncommon. For example, a sample of 100 New
Zealanders were surveyed and almost 10% admitted to moderate
to extreme driving fear and anxiety about driving, where women
reported more negative emotions related to driving than did men
(Taylor and Paki, 2008). In fact, anxiety can be construed in a number of ways as pertains to driving. For example, one study found
fewer participants passed a drivers license test when a second testee was present during the test, than when participants were tested
alone (Rosenbloom et al., 2007). This speaks to a fear of social observation and/or evaluation, a form of state anxiety that is common to
many people, but likely heightened for those with higher levels of
trait anxiety.

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

Recently, Shahar (2009) studied trait anxiety in a sample of

120 Israeli male drivers, and found that drivers with higher levels of anxiety engaged in riskier behaviors. This was interpreted
as being a function of worry as a limiting factor on overall working memory, limiting cognitive capacity that could otherwise be
used for driving tasks. It was also determined that drivers with
higher anxiety exhibited more aggressive behaviors, which was
attributed to poorer emotional adjustment. On the other hand, a
study of over 1000 Australian young adult drivers found no connection between anxiety and risky diving (Vassallo et al., 2008). While
sample types and methods used to assess anxiety (the State-Trait
Anxiety Inventory versus the Depression Anxiety Stress Scales and
Revised Manifest Anxiety Scale, respectively) differed, discrepant
ndings such as these highlight a need for further research.
The current study examined the relationship between selfreported anxiety and dangerous driving behaviors. Dula and
Geller (2004) suggested Dangerous Driving be used as a primary trafc safety research term, and that this main category be
divided into subdivisions of Aggressive, Risky, and Negative Cognitive/Emotional Driving. The theoretical importance of such distinctions seems clear. Road rage implies aggression, but researchers
have used that term and driver aggression to mean varied things.
True aggression requires intent to harm to be present, and many
previous denitions of driver aggression did not meet this criterion.
While aggression is often thought of as a physical act, it is possible for one to intend to harm in a non-physical manner as is done
with insults or gestures intended to intimidate, insult or otherwise
make another feel bad. Rarely, intent to harm may be present without negative emotions on the part of the aggressor, such as in the
case of a psychopath or a rufan taking pleasure in intimidating
others or showing off.
Rarely do such behaviors as red light running, speeding, or weaving in and out of trafc, reect intent to harm someone. Nor are such
behaviors necessarily associated with negative emotions or cognitions. These behaviors are better classied as risky rather than
aggressive or negative cognitive/emotional, but they are nonetheless dangerous.
Negative cognitive and/or emotional driving also warrants separate consideration. A driver might become angry or upset at another
driver but not actually aggress toward the target of that anger. An
upset driver may not willfully engage in particularly risky driving behaviors. However, being cognitively preoccupied with the
feelings of anger and/or of being offended, takes away from the
cognitive resources that could otherwise be applied to the driving
situation. Of course, when a driver experiences what the public or
media term road rage, s/he then is likely to exhibit behaviors in all
three domains of dangerous driving simultaneously.
Previous literature (e.g., Ehring et al., 2008) suggested that
drivers with travel phobia tend to employ safer driving behaviors (e.g., drive more slowly in general, check their mirrors more
frequently). However, this study is concerned with general anxiety symptoms in and not travel phobia specically (nor any other
type of specic anxiety disorder). As anxiety consumes cognitive resources (Eysenck and Byrne, 1992; Gucciardi and Dimmock,
2008) and as safe driving requires sustained attention and emotional composure, it was hypothesized that higher levels of anxiety
would be related more an increased reporting of dangerous driving
behaviors.
2. Method
2.1. Instruments
2.1.1. Dula Dangerous Driving Index (3DI)
The 3DI was created to measure dangerous driving behaviors
in three subcategories: Aggressive Driving (AD, 7 items), Negative-

2051

Emotional Driving (NCE, 9 items), and Risky Driving (RD, 11 items)

(Dula and Ballard, 2003). Dangerous Driving Total (DDT) scores are
derived by summing the 28 items which are scored on a Likert scale
with 1 = Never, 2 = Rarely, 3 = Sometimes, 4 = Often, and 5 = Always.
Subscale scores are calculated by adding the items within each
scale and alpha coefcients for all 3DI scales have ranged from
.73 to .92. Evidence for concurrent, divergent, and predictive validity has been demonstrated (Dula, 2003; Dula and Ballard, 2003)
and a recent cross-cultural conrmatory factor analytic project
rmly established the theoretical legitimacy of the subscale distinctions (Willemsen et al., 2008). One of the studies in this work
was conducted in Belgium, where the 3DI was translated into Dutch
following the procedure described by Brislin (1980). In addition to
support for the factor distinctions, the results showed the 3DI had
comparable reliability and validity across cultures. Internal consistency for each 3DI scale in the present sample was as follows: AD
= .85; NCE = .83; RD = .85; and DDT = .93.
Though being only comprised of two items, a 3DI Drunk Driving
(DD) factor has been identied and has shown evidence of validity,
with prior alpha coefcients ranging from .67 to .79 (Willemsen
et al., 2008). The RD subscale is inclusive of these two items, as
driving intoxicated per se is a form of risky as opposed to aggressive
and/or negative cognitive/emotional driving. Even so, these two
items were added together in the present study and are included
for inspection as a separate subscale factor. Note, however, that
the RD subscale as presented below still includes these two items,
consistent with Ballard and Dula (2003). Similarly, these two items
are not counted twice in DDT scores. The DD scale showed adequate
internal consistency in the present sample, with = .77.
2.1.2. Propensity for Angry Driving Scale (PADS; Depasquale et
al., 2001)
The PADS consists of 19 scored hypothetical driving situations
(e.g., You are driving on a single lane road. For no apparent reason, the car in front of you is constantly braking and accelerating,
causing you to drive in the same manner. How do you respond?).
After reading the prompt scenario, participants select one of four
responses, weighted for relative hostility (e.g., Slow down a little
and keep a safe distance; Deliberately tailgate the car and occasionally lay on the horn). When rst developed, the PADS had excellent
internal consistency, with = 0.89, and a four-week testretest reliability of r = 0.91 (Depasquale et al., 2001). The utility, reliability,
and validity of the measure has been conrmed in other studies,
including with use in cross-cultural samples in Britain and Australia (Leal and Pachana, 2008; Brookings et al., 2008; Maxwell et
al., 2005; Dahlen and Ragan, 2004). In the present sample, the PADS
showed sufcient internal consistency, with = .81.
2.1.3. Beck Anxiety Inventory (BAI, Beck et al., 1988)
From a continuum perspective, the BAI is a distinguished measure of the presence of current anxiety in a persons life. The BAI
has been a popular clinical and research measure for many years
due to its high levels of internal consistency, its brevity (21 items),
and because it was designed to distinguished anxiety from depression symptoms (Steer et al., 1993). A review of the BAIs use in the
literature showed alpha coefcients are generally reported at .83
or better. While testretest coefcients are reported from .35 to
.83, this variability is likely due to the wide range of time intervals
used (between 1 and 16 weeks) in these studies, where the BAI
is only designed to measure symptoms experienced in past week
(de Ayala et al., 2005). Symptoms are rated on a four-point Likert scale (0 = Not At All, 1 = Mildly, 2 = Moderately, and 3 = Severely)
and scores are derived from adding the values of all responses. In
the present sample, the BAI demonstrated excellent internal consistency, with = .93.

2052

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

Hewitt and Norton (1993) found that though women had a

higher mean of reported anxiety, factor structure similarities suggested the nature of anxiety symptoms is comparable between
men and women. Even then it was noted differing factor structures
had been identied, ranging from two (somatic and cognitive) to
ve. More recently Leyfer, Ruberg, and Woodruff-Borden (2005)
recognized a four factor model: subjective, autonomic, neurophysiological, and panic. They found the BAI to be strongest in its ability
to assess symptoms of panic more than any other specic disorder.
Manne et al. (2001) found the BAI was not predictive of a specic
anxiety disorder, but noted this could be due to its brevity or its
nature as a general measure. This latter quality, along with its wellestablished reliability and validity, was the impetus for its being
chosen for the present study.

2.1.4. Demographic questionnaire

A general survey was used to gather self-reported information on age, gender, ethnicity, and relevant trafc safety variables
which included total number of: crashes caused in the past three
years; times having driven intoxicated in the past year; moving violation citations received in the past ve years; seat belt
citations received in the past ve years; as well as a variety
of additional dangerous driving indicators. On these latter variables, participants were asked to estimate the number of times
they engaged in a behavior or experienced an event in the last
two weeks. These included the following: speeding in excess
of 10 miles per hour above the posted speed limit; running red
lights/stop signs; tailgating; weaving in and out of trafc; cutting off another driver in trafc; yelling or cursing at another
driver (such that the other driver might have heard you or
seen you and know you were yelling/cursing them); making
an obscene gesture at another driver; and being honked at by
another driver.
Participants interpreted for themselves the meaning of driven
while intoxicated, tailgated another vehicle, weaved in/out
of trafc, cut off another vehicle, and obscene gesture, rather
than having strict operational denitions for these behaviors. This
allowed participants to assess whether they felt their own behavior rose to a level signicant enough to warrant self-applying
these labels. As estimations of ones various behavioral frequencies across a two-week period are surely inexact, this method
assumes the behaviors in question occur with greater or lesser
frequency in actuality. However imprecise, it is reasonable to presuppose that those who engage in various behaviors in greater
frequency in reality, will report higher levels of the behaviors
than those who actually engage in these behaviors with lesser
frequency. For example, a driver who in actuality tailgates 25
times per week on average, might estimate s/he does so 25 times
every two weeks. On the other hand, a driver who actually tailgates 25 times per week on average, might estimate his/her
rate as 15 times every two weeks. While lacking in ideal levels of reliability, this approach produces a face valid gauge of
greater and lesser levels of dangerous driving nonetheless. This
has been conrmed by consistent convergent correlations of said
behaviors with scores on driving measures that are denitively
reliable and valid (e.g., Dula and Ballard, 2003; Willemsen et al.,
2008).
As there were many single-item variables (13) assessed, a composite was created to yield a more reliable gauge of dangerous
driving propensity, based on these behavioral frequency estimations. The total numbers reported on all single-items were all
summed and averaged to produce a combined variable as a global
approximation of total riskiness as a driver. These 13 items (Mean13) provided yet another measure with which to compare the BAI,
3DI, and PADS.

2.2. Participants/Procedure
There were 1121 participants who completed the 3DI, PADS, and
BAI, including 756 females (67.4%) and 365 males (32.6%). Age was
reported by 1079 (96.3%) with an average age of 21.34 (SD = 5.61;
ranging from 17 to 55). A total of 966 (86.2%) reported being
European-American/White; 29 (2.6%) reported being AfricanAmerican/Black; 15 (1.3%) reported being American Indian; 24
(2.1%) reported being Asian-American; 10 (0.9%) reported being
Hispanic-American; 18 (1.6%) reported being a Citizen of a Foreign Country; 39 (3.5%) reported being Other (than the available
categories); and 20 (1.8%) did not report their race/ethnicity. A
total of 1054 (94.0%) reported the number of years they had been
driving, which averaged 5.43 years (SD = 5.33; ranging from 0.5 to
35). A total of 1082 (96.5%) reported the total number of crashes
that occurred while they were the driver in the past 3 years, with
a mean of 0.53 (SD = 0.89; ranging from 0 to 8). A total of 1105
(98.6%) participants reported the total number of trafc citations
they had received in the previous 5 years, where the average was
0.95 (SD = 1.56; ranging from 0 to 15).
Participants took the surveys online and received modest class
credit for doing so. Measures were presented in random order to
avoid sequential presentation and/or fatigue effects. Participants
were placed into three groups based on a quartile split of their
Beck Anxiety Inventory scores, where the overall sample mean was
13.79 (SD = 10.9, ranging from 0 to 56). The three groups were: Low
Anxiety, including 280 participants (25.0% of the sample) with 151
females (53.9% of this group) and 129 males (46.1% of this group),
scoring between 0 and 5; Medium Anxiety, with 559 participants
(49.9% of the sample) of which 403 were females (72.1% of this
group) and 156 were males (27.9% of this group) scoring between
6 and 20; and High Anxiety which held 282 participants (25.1% of
the sample) with 202 females (71.6% of this group) and 80 males
(23.4% of this group) scoring above 20.
3. Results
An independent samples t-test revealed signicant differences
between males and females on the BAI, with the former having a
mean score of 12.45 (SD = 10.93) and the latter a mean of 14.44
(SD = 10.84). The means and standard deviations of all dependent
variables for sex are presented below in Table 1, and the same are
presented for anxiety groups in Table 2.
The BAI, 3DI, and PADS scales were all signicantly and
positively correlated, to varying degrees, with 3DI inter-scale correlations replicating previous ndings (Dula, 2003; Dula and Ballard,
2003; Willemsen et al., 2008; see Table 3 for full correlation matrix).
Previous 3DI and PADS correlations were largely replicated, with
somewhat smaller coefcient magnitudes in the present versus the
earlier study (Dula and Ballard, 2003). Additionally, the Mean-13
measure was signicantly and positively correlated with all BAI,
3DI, and PADS scales, and it is included in Table 3.
3.1. Interaction between sex and anxiety
To preserve the Type 1 error rate given the relatively large number of dependent variables, a 2 3 Factorial Multivariate Analysis
of Variance (MANOVA) procedure was used where independent
variables consisted of levels of sex (male, female) and anxiety (as
measured by the BAI, grouped as low, medium, high based on a
quartile split), and dependent variables consisted of self-reported
dangerous driving variables. Using Wilks Lambda for the combined dependent variables main effects were found for sex (F = 6.62,
p < .001) and for anxiety group (F = 4.12, p < .001), but the interaction between sex and anxiety group only approached signicance

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

Table 1
Means and standard deviations for all dependent variables by sex.
Dependent variables

3DI RD1
3DI NCE
3DI AD1
3DI DD1
3DI Total1
PADS1
At-Fault Crashes (3 yrs)
DUI Episodes (1 yr)1
Tickets Received (5 yrs)1
Seatbelt Citations (5 yrs)1
Speeding 10+ MPH (2 wks)1
Red Lights Run (2 wks)1
Stop Signs Run (2 wks)
Tailgated (2 wks)
Weaved (2 wks)
Cut Off Other (2 wks)
Cursed/Yelled (2 wks)
Obscene Gesture (2 wks)1
Was Honked At (2 wks)
Mean-131

Male

Female

All

Mean

SD

Mean

SD

Mean

SD

23.56
25.11
15.90
3.07
64.58
45.17
0.34
2.80
1.14
0.68
8.43
0.83
0.66
1.56
2.62
0.90
1.40
1.07
0.75
1.78

7.27
6.08
5.76
1.59
16.83
14.75
0.82
13.31
1.71
1.39
16.72
3.72
1.97
3.58
7.78
1.72
4.26
4.09
2.25
2.66

20.89
25.14
14.40
2.74
60.43
40.49
0.32
1.13
0.81
0.50
5.43
0.44
0.51
1.58
2.12
0.92
1.20
0.58
0.61
1.24

6.39
5.99
5.44
1.40
15.63
12.68
0.65
4.84
1.34
1.14
9.51
1.12
1.45
2.96
3.94
2.58
3.05
2.41
2.14
1.52

21.78
25.13
14.90
2.85
61.81
42.05
0.32
1.69
0.92
0.56
6.42
0.58
0.56
1.57
2.29
0.92
1.27
0.74
0.66
1.42

6.81
6.02
5.59
1.47
16.15
13.58
0.71
8.66
1.48
1.23
12.45
2.34
1.64
3.18
5.52
2.33
3.50
3.08
2.17
1.99

Note: 3DI RD = Dula Dangerous Driving Index-Risky Driving; 3DI NCE = Negative
Cognitive/Emotional Driving; 3DI AD = Aggressive Driving; 3DI Total = Dangerous
Driving Total Score; PADS = Propensity for Angry Driving Survey; Mean-13 = average
of all 13 single-item dangerous driving indicators. 1 = signicant difference between
males and females, see Table 4 for F and p-values.

(F = 1.35, p = .10). Therefore, the results presented below are focused

on relations between sex and dangerous driving variables, and
between anxiety levels and the same.
3.2. Main effect of sex
The MANOVA test revealed signicant differences between
males and females on the following dependent variables: 3DI RD,
3DI AD, 3DI DD, 3DI Total, PADS, Driving Intoxicated in the Previous

2053

Year, Moving Vehicle and Seat Belt Citations Received in the Previous Five Years, as well as Speeding Over 10 MPH, Red Lights Run,
and Making an Obscene Gesture Toward Another Driver in the Past
Two Weeks. These latter single-item sex differences contributed
to a signicant difference in males and females on the Mean-13
measure of which they are contributing elements. The Partial Eta
Squared statistic was calculated to estimate main effect sizes, and
they were relatively small for sex (2 = .10). See Table 4 for all F and
p-values and 2p effect sizes on all dependent variables as pertains
to sex differences.

3.3. Main effect of anxiety group

Results from the 2 3 Factorial MANOVA revealed signicant
differences between low, medium and high anxiety groups on all
but the following dependent variables: Moving Vehicle Citations
Received in the Previous Five Years (this approached signicance,
with p = .09), as well as Speeding Over 10 MPH, Weaving in Trafc, and Cutting Off Another Driver in the Past Two Weeks. Main
effect sizes were also relatively small for anxiety (2 = .07). However, given the magnitude of the problem of dangerous driving,
with tens of thousands killed, millions injured, and billions lost due
to crashes every year, reliably accounting for any variance, however
small, is important.
Tukeys HSD Post Hoc Tests were performed to determine signicant differences between anxiety groups. In the following list
of results, groups not specically mentioned in conjunction with
one another were not signicantly different from one another.
Again, means and standard deviations for all dependent variables
can be found in Table 2. The Low Anxiety and the Medium Anxiety groups scored signicantly lower than the High Anxiety group
at the p < .05 level or lower, but not signicantly differently from
one another, on the following variables: 3DI RD, 3DI AD, 3DI DD,
3DI Total (though the difference between Low and Medium groups
approached signicance on this variable, at p = .08), PADS, Driving
Intoxicated in the Previous Year, Stop Signs Run in the Past Two

Table 2
Means and standard deviations for all dependent variables by anxiety groups.
Dependent variables

Anxiety groups
Low anxiety

3DI RD
3DI NCE
3DI AD1
3DI DD1
3DI Total1
PADS1
At-Fault Crashes (3 yrs)3
DUI Episodes (1 yr)1
Tickets Received (5 yrs)
Seatbelt Citations (5 yrs)2
Speeding 10+ MPH (2 wks)
Red Lights Run (2 wks)4
Stop Signs Run (2 wks)1
Tailgated (2 wks)3
Weaved (2 wks)
Cut Off Other (2 wks)2
Cursed/Yelled (2 wks)1
Obscene Gesture (2 wks)1
Was Honked At (2 wks)1
Mean-131

Med. anxiety

High anxiety

All

Mean

SD

Mean

SD

Mean

SD

Mean

SD

20.71
23.13
13.86
2.74
57.70
40.79
0.22
0.95
0.86
0.52
6.39
0.81
0.43
1.23
2.02
0.96
0.82
0.54
0.53
1.25

6.39
6.41
5.61
1.36
16.55
14.39
0.53
3.99
1.54
1.10
13.08
4.21
1.39
3.12
6.98
3.75
2.12
1.54
1.60
1.89

20.93
25.01
14.29
2.68
60.25
40.13
0.33
1.29
0.88
0.50
6.15
0.38
0.49
1.55
2.12
0.75
1.05
0.57
0.52
1.27

6.06
5.44
5.16
1.30
14.47
11.60
0.68
5.84
1.35
1.17
11.66
0.97
1.70
3.10
4.73
1.41
3.60
3.07
1.71
1.86

24.54
27.40
17.17
3.29
69.12
47.14
0.42
3.21
1.08
0.73
7.00
0.70
0.83
1.96
2.88
1.20
2.15
1.28
1.06
1.88

7.83
5.97
5.81
1.79
16.65
15.09
0.90
14.63
1.65
1.46
13.33
1.40
1.74
3.36
5.31
1.87
4.19
4.07
3.21
2.23

21.78
25.13
14.90
2.85
61.81
42.05
0.32
1.69
0.92
0.56
6.42
0.58
0.56
1.57
2.29
0.92
1.27
0.74
0.66
1.42

6.81
6.02
5.59
1.47
16.15
13.58
0.71
8.66
1.48
1.23
12.45
2.34
1.64
3.18
5.52
2.33
3.50
3.08
2.17
1.99

Note: 3DI RD = Dula Dangerous Driving Index-Risky Driving; 3DI NCE = Negative Cognitive/Emotional Driving; 3DI AD = Aggressive Driving; 3DI DD = Drunk Driving; 3DI
Total = Dangerous Driving Total Score; PADS = Propensity for Angry Driving Survey; Mean-13 = average of all 13 single-item dangerous driving indicators. 1 = signicant
difference between Low and Medium Anxiety groups and High Anxiety groups, but not between Low and Medium groups; 2 = signicant differences between Medium and
High Anxiety groups, but not between Low and High Anxiety groups nor between Low and Medium Anxiety groups; 3 = signicant differences between Low and High Anxiety
groups, but not between Low and Medium Anxiety groups nor between Medium and High Anxiety groups; 4 = signicant differences between Low and Medium Anxiety
groups, but not between High and Low Anxiety groups nor between High and Medium Anxiety groups.

2054

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

Table 3
Pearson correlation matrix among Beck Anxiety Inventory, Dula Dangerous Driving Index, and Propensity for Angry Driving Scale, and Mean-13 dangerous driving indicators.
3DI AD
BAI
3DI AD
3DI NCE
3DI RD
3DI DD
3DI DDT
PADS
Mean-13

**

.27

3DI NCE
**

.27
.71**

3DI RD
**

3DI DD
**

.27
.67**
.60**

.20
.36**
.26**
.62**

3DI DDT

PADS

**

Mean-13

**

.31
.89**
.87**
.88**
.48**

.14**
.34**
.33**
.43**
.33**
.42**
.33**

.23
.67**
.54**
.56**
.35**
.67**

Note: ** p < 0.01; BAI = Beck Anxiety Inventory; 3DI RD = Dula Dangerous Driving Index-Risky Driving; 3DI NCE = Negative Cognitive/Emotional Driving; 3DI AD = Aggressive
Driving; 3DI DD = Drunk Driving; 3DI Total = Dangerous Driving Total Score; PADS = Propensity for Angry Driving Survey; Mean-13 = average of all 13 single-item dangerous
driving indicators.

Weeks, Yelling/Cursing At Another Driver in the Past Two Weeks,

Making an Obscene Gesture Toward Another Driver in the Past Two
Weeks, Being Honked At By Another Driver in the Past Two Weeks,
and on the aggregate Mean-13 measure.
High Anxiety group drivers scored signicantly higher (at the
p < .05 level or lower) than the Medium Anxiety group on Seat
Belt Citations in the Previous Five Years and Cutting Off Another
Driver in the Past Two Weeks, and higher than the Low Anxiety
group on At-Fault Crashes in the Previous Three Years and Tailgating Another Vehicle in the Past Two Weeks. The Low and Medium
Anxiety groups scored signicantly differently from one another
on Red Lights Run in the Past Two Weeks (p < .05). There were no
signicant differences between anxiety groups on Moving Violation Citations in the Previous Five Years, Speeding in the Last Two
Weeks, and Weaving in Trafc in the Last Two Weeks. See Table 5
for all F and p-values and 2p effect sizes on all dependent variables
as pertains to anxiety group differences.
4. Discussion
In the current sample, there was no main effect for the interaction between sex and anxiety level on dangerous driving variables,
though the interaction approached signicance. With large numbers of participants in all relevant cells, it is unlikely that there was
insufcient power to detect such an interaction. Still, there were
clear-cut differences between the sexes on many dangerous driving variables, replicating a number of past ndings. On the whole,
the amount of variance accounted for by sex on the dependent variables was small. There were more signicant relationships between

anxiety levels and dangerous driving variables than there were with
sex and dangerous driving, and the current data supported the very
broad hypothesis that higher levels of anxiety would be related to
an increased reporting of dangerous driving.
Previous ndings of males scoring signicantly higher than
females on the 3DI Risky Driving (RD), Aggressive Driving (AD),
Drunk Driving (DD) and Total scale as well as on the Propensity for
Angry Driving Scale (PADS), were replicated in the current sample
(Dahlen and Ragan, 2004; Dula, 2003; Dula and Ballard, 2003; Leal
and Pachana, 2008; Maxwell et al., 2005; Willemsen et al., 2008). Of
particular interest in further validating the necessity of separating
negative cognitions and emotions from aggression in the context
of driving, there were no signicant gender differences on the 3DI
Negative Cognitive/Emotional Driving subscale, as demonstrated a
number of times before (e.g., Dula, 2003; Dula and Ballard, 2003;
Willemsen et al., 2008). At rst glance, the signicant difference
between men and women on the PADS in this sample may seem a
contradictory nding, until one examines the content of the PADS
more closely.
While named the Angry Driving Scale, many of the items allow
for overtly aggressive responses. For example, the answer choices
for the prompt, You are driving your car down a two-lane road.
Without warning, another car pulls out in front of you from a parking lot. You had to brake suddenly to avoid hitting it. How do you
respond? are: a) Let out a sigh of relief and drive on; b) Lean out
your window and yell at the other driver; c) Honk your horn to
let the other driver know he/she almost caused an accident; Follow the other car to its destination so you can give them a piece
of your mind. Two of the four options imply intent to harm the

Table 4
Sex group differences and effect sizes on dangerous driving variables.
Dependent variables

Sum of squares

df

Mean square

Sig.

2p

3DI RD
3DI NCE
3DI AD
3DI DD
3DI Total
PADS
At-Fault Crashes (3 yrs)
DUI Episodes (1 yr)
Tickets Received (5 yrs)
Seatbelt Citations (5 yrs)
Speeding 10+ MPH (2 wks)
Red Lights Run (2 wks)
Stop Signs Run (2 wks)
Tailgated (2 wks)
Weaved (2 wks)
Cut Off Other (2 wks)
Cursed/Yelled (2 wks)
Obscene Gesture (2 wks)
Was Honked At (2 wks)
Mean-13

1855.73
6.22
555.45

1.00
1.00
1.00

1855.73
6.22
555.45

44.16
.18
19.15

.00
.67
.00

.04
.00
.02

4780.29
5761.26
.32
1008.10
27.02
7.27
1970.92
33.11
4.47
.31
59.19
2.11
17.57
60.63
4.44
76.03

1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00

4780.29
5761.26
.32
1008.10
27.02
7.27
1970.92
33.11
4.47
.31
59.19
2.11
17.57
60.63
4.44
76.03

20.13
33.75
.63
13.74
12.48
4.80
12.85
6.15
1.67
.03
1.94
.39
1.46
6.47
.95
19.93

.00
.00
.43
.00
.00
.03
.00
.01
.20
.86
.16
.53
.23
.01
.33
.00

.02
.03
.00
.01
.01
.00
.01
.01
.00
.00
.00
.00
.00
.01
.00
.02

Note: 3DI RD = Dula Dangerous Driving Index-Risky Driving; 3DI NCE = Negative Cognitive/Emotional Driving; 3DI AD = Aggressive Driving; 3DI DD = Drunk Driving; 3DI
Total = Dangerous Driving Total Score; PADS = Propensity for Angry Driving Survey; Mean-13 = average of all 13 single-item dangerous driving indicators.

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

2055

Table 5
Anxiety group differences and effect sizes on dangerous driving variables.
Dependent variables

Sum of squares

df

Mean square

Sig.

2p

3DI RD
3DI NCE
3DI AD
3DI DD
3DI Total
PADS
At-Fault Crashes (3 yrs)
DUI Episodes (1 yr)
Tickets Received (5 yrs)
Seatbelt Citations (5 yrs)
Speeding 10+ MPH (2 wks)
Red Lights Run (2 wks)
Stop Signs Run (2 wks)
Tailgated (2 wks)
Weaved (2 wks)
Cut Off Other (2 wks)
Cursed/Yelled (2 wks)
Obscene Gesture (2 wks)
Was Honked At (2 wks)
Mean-13

2790.73
1985.95
1580.16
78.06
18074.72
9061.29
5.14
1329.54
10.52
9.30
107.90
37.72
15.96
73.02
98.12
18.93
245.07
112.25
43.35
77.43

2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00

1395.36
992.98
790.08
39.03
9037.36
4530.65
2.57
664.77
5.26
4.65
53.95
18.86
7.98
36.51
49.06
9.46
122.54
56.13
21.68
38.72

33.20
29.23
27.23
18.77
38.05
26.54
5.10
9.06
2.43
3.07
.35
3.50
2.99
3.62
1.61
1.75
10.21
5.99
4.62
10.15

.00
.00
.00
00
.00
.00
.01
.00
.09
.05
.70
.03
.05
.03
.20
.17
.00
.00
.01
.00

.06
.06
.05
.04
.07
.05
.01
.02
.00
.01
.00
.01
.01
.01
.00
.00
.02
.01
.01
.02

Note: 3DI RD = Dula Dangerous Driving Index-Risky Driving; 3DI NCE = Negative Cognitive/Emotional Driving; 3DI AD = Aggressive Driving; 3DI DD = Drunk Driving; 3DI
Total = Dangerous Driving Total Score; PADS = Propensity for Angry Driving Survey; Mean-13 = average of all 13 single-item dangerous driving indicators.

other driver psychologically. An answer option on another item is

Start driving right on the rear bumper of the other car and lay on
your horn. Yet another item response option is Give the other
driver the nger and purposely slow down to aggravate the driver
behind you. In fact, a total of 15 of 22 items have cursing or yelling
obscenities, making obscene gestures, or using ones car to intimidate another driver as response options, all of which are clearly
aggressive as opposed to simply negative cognitions/emotions as
is distinguished in the Dangerous Driving paradigm. Another four
have yelling as an option. It is well known that males report
higher levels of overt aggression in general (e.g., Dykeman et
al., 1996; Gladue, 1991; Harris, 1994; Volavka, 1999), and tend
to engage in more rude and aggressive behaviors while driving
(e.g., Joint, 1995; Mizell, 1997). While the typical PADS response
choices do represent anger-based behaviors, the conceptual clarity
between anger and aggression is obscured and thus sex differences
appear.
When kept conceptually distinct, it is apparent that men and
women are equally likely to experience negative cognitions and
emotions while driving. This was conrmed recently by Lonczak,
Neighbors, and Donovan (2007) who found no differences in men
and women on the variable of driving anger. On an aside, they
also reported that men had more trafc citations than women,
which was replicated in the present sample. Thus, it is imperative
that operational denitions differentiate between actual aggression (intention to harm) and negative cognitions-emotions, where
the latter may be experienced without intent to harm someone, but
where such an experience is certainly distracting to the driver and
thus presents a danger.
Interestingly, quite possibly because anxiety takes up cognitive
resources that could otherwise be used to focus on driving, all three
anxiety groups scored signicantly differently from one another on
the 3DI RD scale, with each increasing level of anxiety being associated with greater risk-taking while driving. Higher anxiety was
also related to higher scores on the 3DI AD, DD, and Total scales, as
well as the PADS, indicating more dangerousness across the board
for high anxiety drivers.
One intriguing trend was the decreasing number of male participants in the anxiety groups as anxiety levels rose. Where 46.1%
were male in the low anxiety group, the proportion of males
decreased to 27.9% in the medium anxiety group, to a low of 23.4%
in the high anxiety group. Self-selection processes may have been
at play, where more highly anxious males may have been less likely

to participate, which could have skewed the data and account for
the lack of a signicant interaction between sex and anxiety groups
on dangerous driving variables. Alternatively, males may be more
likely to appear disproportionally in the lower and middle anxiety
ranges because they do in fact have lower levels of anxiety as a
group, but where anxiety has the same effect on them while driving that it does for females. However, the lack of an interaction may
also be explained by previous ndings that suggest there may be
no difference in how the two sexes experience anxiety in general
(Hewitt and Norton, 1993). Or, it may be males are more reluctant
to report having anxiety, and thus they have similar, though unreported, levels of anxiety. While more work is warranted in this area,
it seems that while sex certainly has an effect on dangerous driving
variables, anxiety contributes its own independent effects.
The Mean-13 shows the overall trend, in that there is often little
difference between low and medium anxiety groups on the various
dangerous driving variables, with few exceptions, but the high anxiety group tends to be more dangerous on the whole than the other
two groups. That males averaged signicantly higher than females
on the Mean-13 measure was not a surprise, and in fact, this lends
support to the notion that this assessment has more than just face
validity. However, the reliability of the Mean-13 was poor, as indicated by its almost unacceptable alpha coefcient of .57. This low
level of reliability suggests it is very difcult to precisely recall a
set of specic numbers of behaviors occurring across a two-week
period. Thus, while may of these single items are of some importance (e.g., number of crashes caused, DUI episodes engaged in),
psychometrically sound standardized self-report measures such as
the 3DI and PADS are absolutely essential to the study of dangerous
driving.
An additional caveat includes the inherent limitations of selfreported data (e.g., social desirability, participant error), though the
use of self-report is standard in research on all manner of human
behavior. To further explore the relationship between anxiety and
dangerous driving, it would be useful to examine more objective
data in the future, such as physiological measures of anxiety, simulated driving behavior, actual video recorded driving, and state
driving and/or insurance company records. It bears repeating that
this was a relatively homogenous sample, made up only of college
students, most of whom were Caucasian and female. It would be
helpful to conduct similar studies with more diverse samples of
drivers, to see if results are replicated for different groups and to
enhance generalizability.

2056

C.S. Dula et al. / Accident Analysis and Prevention 42 (2010) 20502056

5. Conclusion
Independent of sex, higher levels of anxiety are associated with
greater levels of dangerous driving as measured by a variety of indicators. Because of their severity, some of these indicators justify
especially close examination, such as DUI and crashes caused. In
addition to reporting signicantly more DUI episodes in the previous year than the low and medium anxiety groups, the high anxiety
group also had signicantly more at-fault crashes than the low anxiety group in the previous three years. This is an important nding,
because despite the differences in sex on so many dangerous driving variables, there was no difference between males and females
on crashes they reported having caused. But there was such a difference between the highest anxiety and lowest anxiety groups. It is
unlikely that the anxiety measured is a result of the greater number
of crashes, as the surveys were lled out on computers away from
driving situations, and as the Beck Anxiety Inventory measures
anxiety symptoms in general, and not driving phobia specically.
The discovery of consistent relationships between higher levels of
general anxiety and higher reported propensities to engage in dangerous driving behaviors is compelling. Moreover, as crashes are
the most serious potential consequences of dangerous driving, and
as all our trafc safety laws are directed toward preventing crashes,
the nding that highly anxious drivers cause more crashes than less
anxious drivers endorses the notion that there is a great need for
additional research and the development of potential interventions
for highly anxious drivers.
References
Alphers, G.W., Wilhelm, F.H., Roth, W.T., 2005. Psychophysiological assessment during exposure in driving phobic patients. Journal of Abnormal Psychology 114 (1),
126139.
American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental
Disorders (4th ed., text revision). American Psychiatric Association, Washington,
DC.
Beck, J.G., Coffey, S.F., 2007. Assessment and treatment of posttraumatic stress disorder after a motor vehicle collision: empirical ndings and clinical observations.
Professional Psychology: Research and Practice 38 (6), 629639.
Beck, A.T., Epstein, N., Brown, G., Steer, R.A., 1988. An inventory for measuring clinical
anxiety: psychometric properties. Journal of Counseling and Clinical Psychology
56 (6), 893897.
Brislin, R.W., 1980. Translation and content analysis of oral and written material. In:
Triandis, H.C., Berry, J.W. (Eds.), Handbook of Cross-cultural Psychology, vol. 1.
Allyn & Bacon, Boston, pp. 389444.
Brookings, J., de Roo, H., Grimone, J., 2008. Predicting driving anger from trait aggression and self-control. Psychological Reports 103, 622624.
Dahlen, E., Ragan, K., 2004. Validation of the Propensity for Angry Driving Scale.
Journal of Safety Research 35, 557563.
de Ayala, R., Vonderharr-Carlson, D., Kim, D., 2005. Assessing the reliability of the
beck anxiety inventory scores. Educational and Psychological Measurement 65,
742756.
Delahanty, D.L., et al., 1997. Acute and chronic distress and posttraumatic stress disorder as a function of responsibility of serious motor vehicle accidents. Journal
of Consulting and Clinical Psychology 65 (4), 560567.
Depasquale, J.P., Geller, E.S., Clarke, S.W., Littleton, L.C., 2001. Measuring road
ragedevelopment of the propensity for angry driving scale. Journal of Safety
Research 32, 116.
Dula, C.S., 2003. Validity and reliability assessment of a dangerous driving self-report
measure. Dissertation Abstracts International, 64, 960B. Document retrievable from the World Wide Web: http://scholar.lib.vt.edu/theses/available/
etd-04032003-171656/unrestricted/Dissertation2.pdf.
Dula, C., Ballard, M.E., 2003. Development and evaluation of a measure of dangerous, aggressive, negative emotional, and risky driving. Journal of Applied Social
Psychology 33, 263282.
Dula, C.S., Geller, E.S., 2004. Risky, aggressive, or emotional driving: addressing the
need for consistent communication in research. Journal of Safety Research 34,
559566.

Dykeman, C., Daehlin, W., Doyle, S., Flamer, H., 1996. Psychological predictors of
school-based violence: implications for school counselors. School Counselor 44,
3547.
Ehlers, A., Hofmann, S., Herda, C., Roth, W., 1994. Clinical characteristics of driving
phobia. Journal of Anxiety Disorders 8, 323339.
Ehlers, A., Mayou, R.A., Bryant, B., 1998. Psychological predictors of chronic posttraumatic stress disorder after motor vehicle accidents. Journal of Abnormal
Psychology 107 (3), 508519.
Ehring, T., Ehlers, A., Glucksman, E., 2008. Do cognitive models help in predicting the
severity of posttraumatic stress disorder, phobia, and depression after motor
vehicle accidents? A prospective longitudinal study. Journal of Consulting and
Clinical Psychology 76 (2), 219230.
Eysenck, M., Byrne, A., 1992. Anxiety and susceptibility to distraction. Personality
and Individual Differences 13, 793798.
Gladue, B.A, 1991. Aggressive behavioral characteristics, hormones, and sexual orientation in men and women. Aggressive Behavior 17, 313326.
Gucciardi, D., Dimmock, J., 2008. Choking under pressure in sensorimotor skills:
Conscious processing or depleted attentional resources? Psychology of Sport
and Exercise 9, 4559.
Harris, M.B., 1994. Gender of a subject and target as mediators of aggression. Journal
of Applied Social Psychology 24, 453471.
Hewitt, P.L., Norton, G.R., 1993. The Beck Anxiety Inventory: a psychometric analysis.
Psychological Assessment 5 (4), 408412.
Hodge, J., 1971. The whiplash neurosis. Psychosomatics: Journal of Consultation
Liaison Psychiatry 12, 245249.
Joint, M. (1995). Road rage. AAA Foundation for Trafc Safety. Retrieved
August 15, 2009, from the World Wide Web: http://www.aaafoundation.org/
resources/index.cfm?button=agdrtext#Road Rage.
Kessler, R.C., Sonnega, A., Bromet, E., Hughes, M., Nelson, C.B., 1995. Posttraumatic
stress disorder in the national comorbidity survey. Archives of General Psychiatry 52, 10481060.
Leal, N., Pachana, N., 2008. Adapting the Propensity for Angry Driving Scale for use
in Australian research. Accident Analysis and Prevention 40, 20082014.
Linnell, M., Easton, S., 2004. The relationship between phobic travel anxiety and
the physical symptoms of whiplash injury. Rehabilitation Psychology 49 (4),
317320.
Llobet, P. (2009). Group hypnotherapy in the treatment of specic phobias: An
exploratory study. Dissertation Abstracts International, 69.
Lonczak, H., Neighbors, C., Donovan, D., 2007. Predicting risky and angry driving as
a function of gender. Accident Analysis & Prevention 39, 536545.
Manne, S., Nereo, N., DuHamel, K., Ostroff, J., Parsons, S., Martini, R., 2001. Anxiety and depression of mothers of children undergoing bone marrow transplant:
symptom prevalence and use of the beck depression and beck anxiety inventories as screening instruments. Journal of Consulting and Clinical Psychology 69
(6), 10371047.
Maxwell, J., Grant, S., Lipkin, S., 2005. Further validation of the propensity for
angry driving scale in British drivers. Personality and Individual Differences 38,
213224.
Mizell, L., 1997. Aggressive driving. AAA Foundation for Trafc Safety. Retrieved
August 15, 2009, from the World Wide Web: http://www.aaafoundation.org/
resources/index.cfm?button=agdrtext#Aggressive Driving.
Mowrer, O. (1960). Learning theory and behavior. Hoboken, NJ US: John Wiley &
Sons Inc.
Rosenbloom, T., Shahar, A., Perlman, A., Estreich, D., Kirzner, E., 2007. Success on a
practical drivers license test with and without the presence of another testee.
Accident Analysis & Prevention 39, 12961301.
Shahar, A., 2009. Self-reported driving behaviors as a function of trait anxiety. Accident Analysis and Prevention 41, 241245.
Steer, R.A., Ranieri, W.F., Beck, A.T., Clark, D.A., 1993. Further evidence for the validity
of the Beck Anxiety Inventory with psychiatric outpatients. Journal of Anxiety
Disorders 7, 195205.
Taylor, J.E., Deane, F.P., Podd, J.V., 2000. Determining the focus of driving fears.
Journal of Anxiety Disorders 14 (5), 453470.
Taylor, J., Paki, D., 2008. Wanna drive? Driving anxiety and fear in a New Zealand
community sample. New Zealand Journal of Psychology 37, 3137.
Vassallo, S., Smart, D., Sanson, A., Cockeld, S., Harris, A., McIntyre, A., et al., 2008.
Risky driving among young Australian drivers II: co-occurrence with other problem behaviours. Accident Analysis & Prevention 40, 376386.
Volavka, J., 1999. The neurobiology of violence: an update. Journal of Neuropsychiatry and Clinical Neurosciences 11, 307314.
Willemsen, J., Dula, C.S., Declercq, F., Verhaeghe, P., 2008. The Dula Dangerous Driving Index: an investigation of reliability and validity across cultures. Accident
Analysis and Prevention 40, 798806.
Williams, S.L., Dooseman, G., Kleield, E., 1984. Comparative effectiveness of guided
mastery and exposure treatments for intractable phobias. Journal of Consulting
and Clinical Psychology 52 (4), 505518.
Wolpe, J., 1958. Psychotherapy by reciprocal inhibition. Stanford University Press,
Oxford England.