P1: GAD

Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

Journal of Psychopathology and Behavioral Assessment, Vol. 26, No. 2, June 2004 (°
C 2004)

Evaluation of Two New Scales Assessing Driving Anger:

The Driving Anger Expression Inventory and the Driver’s
Angry Thoughts Questionnaire

Jerry L. Deffenbacher,1,3 Gail S. White,1 and Rebekah S. Lynch2

Accepted September 9, 2003

This study assessed the validity of the Driving Anger Expression Inventory (DAX) and Driver’s Angry
Thoughts Questionnaire (DATQ). Scales within the DAX and DATQ appeared to assess separate,
correlated constructs. Aggressive forms of thinking and anger expression correlated positively with
each other, trait driving anger, aggressive and risky behavior, some crash-related conditions, general
trait anger, and general forms of anger expression. Positive, constructive forms of thinking and anger
expression correlated positively with each other, but minimally or negatively with other variables.
Specific aggressive forms of thinking and anger expression formed stronger links with each other than
other forms of expression, supporting the discriminant validity of the DATQ and DAX. Hierarchical
regressions with DATQ and DAX scales entered after other variables showed that they added explained
variance above and beyond other measures, supporting the discriminant and incremental validity of
these scales. It was concluded that the DAX and DATQ have utility for researchers and clinicians
interested in angry drivers.

KEY WORDS: angry drivers; anger expression; anger-related thoughts.

Angry, aggressive drivers have received considerable
media attention in the past few years, and public con-
cern about the psychological and health hazard posed by
these drivers has led to increased governmental and police
attention. It would appear that the public’s concerns are
not misplaced. “Road rage,” incidents involving assault
or attempted assault (e.g., firing a gun at another driver
or attempting to run someone off the road) appeared to
increase approximately 7% a year in the United States
during the early 1990s (American Automobile Associa-
tion, 1997). Some public officials have estimated that from
one third to two thirds of injury accidents involve aggres-
sive driving such as speeding and rapid lane changing in
heavy traffic (Martinez, 1997; Snyder, 1997). Addition-
ally, court- and self-referred aggressive drivers experience
1 Department of Psychology, Colorado State University, Fort Collins,
Colorado.
2 Nursing Program, Front Range Community College, Fort Collins,
Colorado.
3 To whom correspondence should be addressed at Department of Psy-
chology, Colorado State University, Fort Collins, Colorado 80523-
1876; e-mail: jld6871@colostate.edu.
higher incidence of intermittent explosive disorder and
other psychopathology (Galovski, Blanchard, & Veazey,
2002), some of which gets played out on the road, and
drivers who physically assault another driver experience
more crashes and traffic tickets (Hemenway & Solnick,
1993).
Highly angry drivers are also at elevated risk
(Deffenbacher, Deffenbacher, Lynch, & Richards, 2003;
Deffenbacher, Filetti, Richards, Lynch, & Oetting, 2003;
Deffenbacher, Huff, Lynch, Oetting, & Salvatore, 2000;
Deffenbacher, Lynch, Filetti, Dahlen, & Oetting, 2003).
High and low anger drivers drive the same number of times
and miles per week, suggesting that they are not exposed to
qualitatively different driving environments which might
account for differences observed. However, more things
on the road provoke high anger driver, they become angry
2.5–3.0 times more often than low anger drivers, and they
become more intensely angry when provoked. High anger
drivers also tend to express their anger in more dysfunc-
tional ways. They are more verbally and physically aggres-
sive and are more likely to use their vehicle as a means of
expressing their anger and are less likely to express anger

87
0882-2689/04/0600-0087/0 °
C 2004 Plenum Publishing Corporation
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004
88
behind the wheel in adaptive and constructive ways. High
anger drivers also engage in 3.5–4.0 times more aggres-
sion and 1.5–2.0 times more risky behavior while driv-
ing and, in some studies, report more close calls, moving
violations, and minor accidents than low anger drivers.
In studies involving driving simulations (Deffenbacher,
Deffenbacher, et al., 2003), high anger drivers, compared
to low anger drivers, engaged in excessive speeding (e.g.,
40% vs. 10% at 10 mph over the limit and 12% vs. 0%
at 20 mph over the limit) in low impedance simulations,
and in high impedance simulations reported more anger
and verbal and physical aggression, engaged in more er-
ratic driving, experienced shorter times and distances to
crashes, and had double the number of crashes. Moreover,
studies of British drivers show that driving anger corre-
lates positively with both aggressive and nonaggressive
driving offenses (Lajunen & Parker, 2001; Lajunen, Park,
& Stradling, 1998; Underwood, Chapman, Wright, &
Crundall, 1999). Other studies also show that high anger
drivers also tend to be more generally angry, impulsive,
and aggressive (Deffenbacher, Deffenbacher, et al., 2003;
Deffenbacher, Filetti, et al., 2003; Lajunen & Parker,
2001). In summary, high anger drivers are more easily, fre-
quently, and intensely angered on the road, tend to express
that anger in aggressive ways, engage in more aggressive
and risky behavior, and experience greater incidence of
some crash-related conditions such as moving violations,
close calls, and the like.
Recently, angry, aggressive drivers have begun to
receive clinical attention (Deffenbacher et al., 2000;
Deffenbacher, Filetti, Lynch, Dahlen, & Oetting, 2002;
Galovski & Blanchard, 2002). As research and practice
move to clinical applications, clinicians and treatment re-
searchers need reliable and valid instruments with which
to assess important characteristics of angry drivers. One
reasonably well validated measure is the Driving Anger
Scale (DAS; Deffenbacher, Oetting, & Lynch, 1994)
which provides a measure of anger intensity across a num-
ber of driving situations. However, other instruments are
needed to screen for and assess issues such as how anger
is expressed behind the wheel and the cognitive elements
of the angry driver’s experience.
Recently, two new self-report instruments, one as-
sessing ways of expressing anger and the other assessing
types of angry cognitions when angered behind the wheel,
have been reported. The Driving Anger Expression In-
ventory (DAX; Deffenbacher, Lynch, Oetting, & Swaim,
2002) provides four, factor analytically derived scales
measuring: (a) Verbal Aggressive Expression involving
expressing anger through things such cursing at or calling
other drivers name, asking negative rhetorical questions
about the person’s driving and character, and the like;
10:51 Style file version June 25th, 2002
Deffenbacher, White, and Lynch
(b) Personal Physical Aggressive Expression in which the
drivers uses his/her physical being to express anger and
intimidate through things such as hostile gestures and get-
ting out of the vehicle to fight with an offending driver; (c)
Using the Vehicle to Express Anger involving behaviors
such as flashing lights, slowing down or speeding up to
frustrate or retaliate upon another driver, or purposefully
tailgating another driver; and (d) Adaptive/Constructive
Expression involves the driver’s attempts to focus on
positive coping and safe driving through things such as
cognitive reframing and relaxing and distracting behav-
iors. The Driver’s Angry Thoughts Questionnaire (DATQ;
Deffenbacher, Petrilli, Lynch, Oetting, & Swaim, 2003)
provides five, factor analytically derived scales assess-
ing the following cognitive characteristics: (a) Judgmental
and Disbelieving Thinking involves thoughts of mild dero-
gation, questioning of other driver’s skills, abilities, and
motivations, statements implying others should not be al-
lowed to drive, and rhetorical questions indicating that
they cannot believe how others are driving; (b) Pejora-
tive Labeling and Verbally Aggressive Thinking involves
highly negative thoughts about the other driver (e.g., name
calling) and thinking about how the person would like to
engage in verbally aggressive behavior (e.g., telling the
person off); (c) Physically Aggressive Thinking involv-
ing thoughts of wanting to hurt others and behaviors they
would engage in to do so (e.g., beating the hell out of
the offending driver); (d) Revenge and Retaliatory Think-
ing involves thoughts of getting back at others and the
behaviors needed to exact revenge (e.g., boxing the other
driver in to show him/her); and (e) Coping Self-instruction
addresses instructing one’s self to engage in positive
coping activities such as relaxing and accepting bad
situations.
To be of value as clinical assessment devices, new
scales such as the DAX and DATQ must meet at least two
basic criteria. First, they must have clinical face validity.
That is, they must appear to assess important parameters
and processes related to the client and presenting prob-
lems. Both scales appear to meet this test. They address
the cognitive and expressive dimensions of anger while
driving. At least on their face, the DAX and DATQ could
extend the clinical interview and other assessment strate-
gies such as self-monitoring to flesh out the cognitive and
expressive elements of anger behind the wheel. Second,
they must correlate in logical and reasonable ways with
other clinically important elements of client experience,
such as intensity of anger, aggression, and risky behavior
while driving. Initial validity work (see Method section
for review) suggests that both scales meet this criterion,
but additional work is needed to strengthen confidence
in findings. The value of the DAX and DATQ would be
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578
Two New Scales for Driving Anger
strengthened further if it could be shown that they provide
information and added predictive power above and beyond
that provided by instruments that are readily available such
as the Driving Anger Scale (DAS; Deffenbacher et al.,
1994) or the Trait Anger Scale (TAS; Spielberger, 1988).
If, for example, the DAX and DATQ provide little infor-
mation beyond that provided by the DAS, then they might
not add much clinical information and would have little
to recommend them as additional outcome measures. If,
on the other hand, the DAX and DATQ add information
and predictive power beyond that provided by other in-
struments, this would support added value to clinicians
and outcome researchers.
This research addressed these issues in three ways.
First, it provided a partial replication of correlational find-
ings from the Deffenbacher, Lynch, et al. (2002) and
Deffenbacher et al. (2003) studies and included the DAX
and DATQ in the same study. Second, it explored the added
value of the DAX and DATQ in hierarchical regression
models in which other variables such as gender and the
DAS were entered first so that the contribution of the DAX
or DATQ above beyond these variables could be explored.
Evidence of incremental validity would support the value
and potential utility of these measures. Third, evidence of
discriminant validity was explored. For example, if spe-
cific scales of the DAX or DATQ were differentially re-
lated to each other or other phenomena such as aggression
or risky behavior on the road, this would suggest spe-
cific relationships which would aid in the understanding of
clients and might provide clinically useful relationships to
target in treatment. For example, if revengeful/retaliatory
thinking was most strongly related to using the vehicle to
express anger and to aggression, then it would suggest
that this cognitive–expressive-behavioral link be targeted
for individuals assessed high on these variables. Discrim-
inant validity was explored through tests of differences of
strengths of correlations and by weightings of variables in
regression models.
METHOD
Paticipants
Participants were 436 (218 male, 218 female) intro-
ductory psychology students (M age = 18.75, SD = 2.53)
at a Western State University of over 25,000 enrollment.
Of these, 70.2% were freshmen, 20.0% sophomores, and
9.8% juniors and seniors, and 68.3% graduated from
high schools in towns over 10,000 in population, 16.3%
in towns from 5,000–10,000, and 15.3% in towns un-
der 5,000. Ethnic breakdown was 82.6% non-Hispanic
White, 4.8% Hispanic/Latino, 1.8% Asian American,
January 28, 2004 10:51 Style file version June 25th, 2002
89
1.4% African American, 1.1% Native American, and 4.6%
of mixed or other background. Participants received one
of four research credits for participation.
Instruments
Demographic Information
On the first page of the packet, participants indicated
their age, gender, and year in school, endorsed all ethnic-
ity categories that applied (i.e., Native American, African
American, Asian American, Hispanic/Latino, White non-
Hispanic, or multiple ethnic background/other), and
checked the size of town from which they graduated
high school (i.e., under 1,000, 1,000–2,000, 2,000–3,000,
3,000–5,000, 5,000–10,000, and over 10,000).
Driving Anger Scale (DAS)
The 14-item DAS (Deffenbacher et al., 1994) mea-
sures the propensity to anger when driving. Respondents
rated on a 5-point scale (1 = not at all, 5 = very much)
the intensity of anger experienced when they encounter
the situation described (e.g., someone is weaving in and
out of traffic or someone runs a red light or stop sign).
Alpha reliabilities range from .80 to .93 and 10-week
test–retest reliability was .84 (Deffenbacher, 2000).
General and clinical samples have shown that driving
anger as measured by the DAS correlates positively
with anger in both the individual’s most provocative
situations and in commonly occurring driving events,
with state anger and aggressiveness during visualization
of provocative driving situations and high impedance
driving simulations, with the frequency and intensity of
anger in day-to-day driving, with aggressive forms of ex-
pressing anger behind the wheel, and with the frequency
of aggressive and risky behavior and some crash-related
outcomes such as moving violations (Deffenbacher et al.,
2000; Deffenbacher, Deffenbacher, et al., 2003; Deffen-
bacher, Filetti, et al., 2003; Deffenbacher, Lynch, et al.,
2003; Deffenbacher, Lynch, Deffenbacher, & Oetting,
2001). Driving anger also correlates positively with
general trait anger, aggressive and less controlled forms
of general anger expression, impulsivity, and trait anxiety
(Deffenbacher et al., 2000; Deffenbacher, Deffenbacher,
et al., 2003; Deffenbacher, Lynch, et al., 2003).
Driving Anger Expression Inventory (DAX)
The 49-item DAX is a self-report of how people ex-
press their anger behind the wheel (Deffenbacher et al.,
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004
90
2001; Deffenbacher, Lynch, et al., 2002). Respondents
rated on a 4-point scale (1 = almost never, 4 = almost
always) how often they express their anger while driving
in the manner described in the item. The DAX yields four
reliable forms of anger expression. The 12-item Verbal
Aggressive Expression Scale (αs = .88–.90) assesses peo-
ple’s tendency to express their driving anger through ver-
bally aggressive means (e.g., yelling or swearing at an-
other driver). The 11-item Personal Physical Aggressive
Expression Scale (αs = .80–.84) assesses the person’s
propensity to use his/her physical presence to express
anger and intimidate (e.g., giving the other driver the finger
or getting out of the car to engage in a fist fight). The 11-
item Use of the Vehicle to Express Anger Scale (αs = .86–
.89) measures the tendency to use the vehicle in aggressive
expression of anger (e.g., speeding up to frustrate another
driver or driving up close to another driver’s bumper). The
fourth scale, the 15-item Adaptive/Constructive Expres-
sion Scale (αs = .89–.90), assesses the person’s tendency
toward positive coping in the face of anger on the road
(e.g., engaging in relaxation or thinking about things to
distract one from provocation). Verbal, personal physical,
and vehicular aggressive forms of driving anger expres-
sion correlate positively with each other (rs = .40–.50),
the DAS, and risky and aggressive behavior on the road,
but are uncorrelated with or form small negative correla-
tions with the adaptive/constructive form of expression.
Adaptive/constructive expression tends to form smaller
negative correlations with driving-related variables. Ver-
bal, personal physical, and vehicular aggressive forms of
anger expression also tend to correlate positively with
aggressive cognitions. Adaptive/constructive expression
also correlated positively with coping self-instructions,
but was less correlated with other cognitions behind the
wheel.
Driver’s Angry Thoughts Questionnaire (DATQ)
On the 65-item DATQ respondents rated on a 5-point
scale (1 = not at all, 5 = all the time) how often they have
the thought listed or one similar to it when they are angry
behind the wheel (Deffenbacher et al., 2003). The DATQ
yields five scales of driving-related cognitions: (a) the 21-
item Judgmental and Disbelieving Thinking (α = .94) as-
sesses mild to moderate derogation of other drivers, state-
ments or questions indicating that they cannot believe that
others are driving the way that they are or that they should
not be allowed to drive (e.g., Who do they think they are
or They are clueless.); (b) the 13-item Pejorative Label-
ing and Verbally Aggressive Thinking (α = .92) involves
much more negative labels and judgments and thoughts of
10:51 Style file version June 25th, 2002
Deffenbacher, White, and Lynch
engaging in verbally aggressive behavior (e.g., What an
ass or I am so pissed.); (c) the 14-item Revenge and Retal-
iatory Thinking (α = .93) assesses the tendency to think
about getting back at others and the behaviors necessary to
do so (e.g., I am not going to let them do that to me or I am
going to teach them a lesson.); (d) the 8-item Physically
Aggressive Thinking assesses thoughts of wanting to hurt
others and thinking about the behaviors necessary to do so
(e.g., I want to run them off the road or I would like to beat
the hell out of them.); and (e) the 9-item (α = .83) Cop-
ing Self-instruction assesses thoughts of positive, adaptive
coping (e.g., Cope with it, sometimes you just have to live
with bad drivers or Just pay attention to my driving, oth-
ers can be crazy if they want.). Judgmental/disbelieving
thinking correlated positively with other forms of think-
ing, whereas pejorative labeling/verbally aggressive, re-
vengeful/retaliatory, and physically aggressive thinking
correlated highly and positively with each other and neg-
atively with coping self-instructions. Forms of angry cog-
nitions behind the wheel generally correlated in expected
directions with trait driving anger, anger in specific driv-
ing situations, aggressive forms of driving anger expres-
sion, aggressive and risky behavior on the road, and some
crash-related conditions. Cognitions in the DATQ were
better predictors of driving-related variables than a gen-
eral measure of hostile cognitions (Deffenbacher et al.,
2003).
Driving Survey
The Driving Survey (Deffenbacher et al., 2000;
Deffenbacher, Deffenbacher, et al., 2003) assessed the
three-month frequencies of 13 aggressive (e.g., mak-
ing an angry gesture at another driver) and 15 risky
(e.g., speeding 20+ mph over the speed limit) behav-
iors and three crash-related conditions (i.e., loss of con-
centration, minor loss of vehicular control, and close
calls while driving) and last year frequencies of mov-
ing violations and minor and major accidents. Partici-
pants circled a number from 0 to 5+ (with 5+ being
treated as 5 in analyses) to indicate the frequency of
the behavior or condition. Three-month reports of Ag-
gression (αs = .85–.89) and Risky Behavior (αs = .83–
.86) form reliable indices, whereas the six crash-related
conditions do not (αs = .41–.55). Crash-related condi-
tions, therefore, were analyzed individually. Compared to
low anger drivers, high anger drivers report more aggres-
sive and risky behavior and more of some crash-related
conditions (Deffenbacher et al., 2000; Deffenbacher,
Deffenbacher, et al., 2003; Deffenbacher, Lynch, et al.,
2002).
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578
Two New Scales for Driving Anger
Trait Anger Scale (TAS)
The 10-item TAS assessed general feelings of anger
(Spielberger, 1988). Participants rated on a 4-point scale
(1 = almost never, 4 = almost always) the degree to which
the statement reflects how he/she typically feels or re-
acts. TAS α reliabilities range from .81 to .91, and 2-week
to 2-month test–retest range reliabilities from .70 to .77
(Jacobs, Latham, & Brown, 1988; Morris, Deffenbacher,
Lynch, & Oetting, 1996). The TAS is a well validated mea-
sure correlating positively with measures of anger and ag-
gression (Deffenbacher et al., 1996; Spielberger, 1988,
1999) and anger consequences (Deffenbacher, Oetting,
Lynch, & Morris, 1996) and correlates more with anger
variables than with other cognitive, behavioral, emotional,
and personality variables (Deffenbacher, Oetting, Lynch,
et al., 1996; Deffenbacher, Oetting, Thwaites, et al., 1996,
Spielberger, 1999).
Anger Expression Inventory (AX)
The 24-item AX was included as a general mea-
sure of anger expression (Spielberger, 1988). The 24-item
AX asked respondents to rate how they generally express
themselves when angry or furious. They rated on a 4-point
scale (1 = almost never, 4 = almost always) how descrip-
tive the item was of the person’s way of expressing anger
generally. The AX is composed of three 8-item scales:
(a) Anger-In assessing the tendency to suppress anger
and harbor grudges; (b) Anger-Out assessing the ten-
dency to express anger outwardly and negatively; and (c)
Anger-Control assessing attempts to calm down and con-
trol anger. AX scale α reliabilities range from .73 to .84.
Anger-In tends to be relatively uncorrelated with Anger-
Out and Anger-In, whereas Anger-Out and Anger-Control
are negatively correlated. Anger-In and Anger-Out cor-
relate positively with the TAS, whereas Anger-Control
correlates negatively with the TAS. Validity for the AX
scales is reflected in different patterns of correlations with
measures of anger, anger consequences, personality, and
physiological reactions (Deffenbacher, Oetting, Thwaites,
et al., 1996; Spielberger, 1988).
Procedure
Students read about the study on the departmental
website which described research opportunities and which
described this research as a one-credit study involving
emotion, especially anger, and driving behavior. Inter-
ested students signed up for a day and time fitting their
schedules. The study was conducted in university class-
rooms holding 100–200 students. Upon arriving, students
January 28, 2004 10:51 Style file version June 25th, 2002
91
received two informed consent forms, one of which they
read, signed, and returned, the other of which they kept
for their records. After signing consent forms, they com-
pleted one of three counterbalanced orders of the DAS,
DAX, DATQ, Driving Survey, TAS, and AX. However,
the TAS and AX always followed driving measures, be-
cause pilot work suggested this reduced confusion about
whether general measures referred to driving or not. When
students finished surveys, they received a written debrief-
ing statement.
RESULTS
Preliminary analyses showed that demographics,
save gender, were not related to findings.
Gender Differences
Gender differences (Table I) on the DAS were as-
sessed by a one-way ANOVA, whereas the DAX, DATQ,
Driving Survey, and general anger measures (TAS and
AX) were analyzed by one-way MANOVAs employing
the Wilks’s λ statistic. Effect sizes are in terms of η2 and are
interpreted according to Cohen’s criteria (Cohen, 1988)
where an effect size of .01 to .04 is considered small, .05
to .14 moderate, and greater than .14 as large. The DAS did
not show gender effects (Table I), but the DAX revealed a
moderate multivariate gender effect, F(4, 427) = 10.68,
p < .001, η2 = 0.09. Men and women did not differ on
verbal expression of anger while driving, but men re-
ported greater physical and vehicular expression and less
adaptive/constructive means of expressing anger (Table I).
Significant gender effects were small to moderate. The
DATQ demonstrated a large multivariate gender effect,
F(5, 426) = 18.84, p < .001, η2 = 0.18, due to men re-
porting more revengeful/retaliatory and physically aggres-
sive thinking than women (Table I). Effect sizes were
moderate and large, respectively. Men and women did not
differ on other anger-related patterns of thinking. Aggres-
sive and Risky Behavior from the Driving Survey revealed
a small multivariate gender effect, F(2, 429) = 5.83, p <
.01, η2 = 0.03, with men reporting more aggressive and
risky behavior. Effect sizes for these two variables were
small. The six crash-related conditions demonstrated a
small multivariate gender effect, F(6, 425) = 2.30, p <
.05, η2 = 0.03. Men and women differed only on mov-
ing violations, due to men reporting more tickets. General
anger measures also revealed a small multivariate gender
effect, F(4, 427) = 3.71, p < .01, η2 = 0.03, due to men
reporting more general trait anger and Anger-Out than
women. Effects sizes for both were small.
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

92 Deffenbacher, White, and Lynch

Table I. Gender Differences on the DAS, DAX, DATQ, Driving Survey, TAS, and AX

Men Women
Measures M SD M SD Univariate gender, F(1, 431) Gender effect, η2

DAS 46.94 9.23 45.80 9.45 1.60 0.00

DAX-verbal 27.12 7.73 26.31 8.25 1.10 0.00
DAX-personal physical 14.83 3.79 13.40 2.26 22.94∗∗∗ 0.05
DAX-use of vehicle 20.43 6.49 18.08 5.58 16.34∗∗∗ 0.04
DAX-adaptive/constructive 31.94 7.93 35.36 8.91 17.80∗∗∗ 0.04
DATQ-JD 54.00 17.85 55.11 20.56 0.35 0.00
DATQ-PLVA 41.15 12.46 38.84 12.87 3.58 0.01
DATQ-RR 29.55 13.64 22.49 9.50 38.90∗∗∗ 0.08
DATQ-PA 15.08 8.46 9.47 3.54 80.72∗∗∗ 0.16
DATQ-CSI 18.10 5.86 18.94 6.82 1.90 0.00
Aggression (3 months) 12.22 11.40 9.85 9.02 5.73∗ 0.01
Risky behavior (3 months) 24.59 16.28 19.85 12.47 11.56∗∗∗ 0.03
Loss of concentration (3 months) 2.58 1.73 2.66 1.70 0.23 0.00
Minor loss of control (3 months) 1.16 1.25 1.10 1.30 0.27 0.00
Close call (3 months) 1.05 1.28 1.21 1.19 2.00 0.01
Moving tickets (last year) 1.47 1.46 1.05 1.29 9.79∗∗ 0.02
Minor accidents (last year) 0.92 1.00 0.87 0.98 0.30 0.00
Major accidents (last year) 0.31 0.64 0.25 0.53 1.04 0.00
TAS 19.61 5.56 18.20 5.09 7.58∗∗ 0.02
Anger-in 16.81 4.47 16.23 4.43 1.80 0.00
Anger-out 16.90 4.00 15.87 3.75 7.74∗∗ 0.02
Anger-control 23.02 4.85 22.87 5.11 0.09 0.00

Note. DAS = Driving Anger Scale; DAX = Driving Anger Expression; DATQ = Driver’s Angry Thoughts Questionnaire; JD =
Judgmental/Disbelieving thinking; PLVA = Pejorative Label/Verbally Aggressive thinking; RR = Revenge and Retaliatory thinking;
PA = Physically Aggressive thinking; CSI = Coping Self-Instruction; and TAS = Trait Anger Scale.
∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

Correlations Between Measures by Gender
Because some gender differences were found in the
distributions of variables, potential gender differences in
correlations were explored by tests for differences in
independent correlations (Bruning & Kintz, 1997). Be-
cause of the number of correlations involved here and in
other analyses α was set a p < .01 to reduce Type I error.
The three aggressive forms of expressing driving
anger (verbal, physical, and use of the vehicle) correlated
positively with each other and with the DAS (Table II).
Adaptive/constructive expression was uncorrelated with
or correlated negatively with other forms of driving anger
expression and the DAS. Relationships were similar for
men and women, but the correlation between verbal
and physical forms of anger expression was stronger
for women, whereas the correlation between physical
and vehicular forms of expression was stronger for
men.
Judgmental/disbelieving thinking correlated posi-
tively with other forms of driving anger-related thoughts
(Table III), and pejorative labeling/verbally aggres-
sive, revengeful/retaliatory, and physically aggressive
thoughts correlated positively with each other. Coping

Table II. Correlations Between Trait Driving Anger (DAS) and Forms of Anger Expression (DAX)

Measure DAX-verbal DAX-physical DAX-vehicular DAX-adaptive/constructive

DAS .49(.35) .40(.30) .47(.37) −.25(−.02)

DAX-verbal .57a (.73)b .53(.54) −.21(−.27)
DAX-physical .67a (.45)b −.27(−.14)
DAX-vehicle −.42(−.38)

Note. Correlations for men are outside the parentheses, correlations for women are inside parentheses.
Different subscripts indicate a significant difference between genders in strengths of correlations ( p < .01).
DAS = Driving Anger Scale; and DAX = Driving Anger Expression. r > .18, p < .01.r > .22, p < .001.
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

Two New Scales for Driving Anger 93

Table III. Correlation Between Trait Driving Anger (DAS) and Angry Thoughts (DATQ)

Measure DATQ-JD DATQ-PLVA DATQ-RR DATQ-PA DATQ-CSI

DAS .32(.36) .52(.43) .46(.35) .40(.19) −.08(.09)

DATQ-JD .62(.70) .43(.48) .34(.25) .35(.47)
DATQ-PLVA .64(.61) .61a (.37)b −.05(.15)
DATQ-RR .75(.63) −.14(.03)
DATQ-PA −.17(.11)

Note. Correlations for men are outside the parentheses, correlations for women are inside parenthe-
ses. Different subscripts indicate a significant difference between genders in strengths of correlations
( p < .01). DAS = Driving Anger Scale; DATQ = Driver’s Angry Thoughts Questionnaire; JD = Judg-
mental/Disbelieving thinking; PLVA = Pejorative Label/Verbally Aggressive thinking; RR = Revenge
and Retaliatory thinking; PA = Physically Aggressive thinking; and CSI = Coping Self-instruction.
r > .18, p < .01. r > .22, p < .001.

self-instructional thoughts correlated only with judgmen-
tal/disbelieving cognitions. Trait driving anger (DAS) cor-
related positively with all forms of anger-related thinking,
except for coping self-instructions. The pattern of corre-
lations for men and women were similar, but for men the
correlation of physically aggressive thinking with pejora-
tive labeling/verbally aggressive thinking was larger than
for women.
The relationships between angry thinking (DATQ)
and anger expression (DAX) are reported in Table IV.
Judgmental/disbelieving, pejorative labeling/verbally ag-
gressive, revengeful/retaliatory, and physically aggressive
thinking correlated positively with verbal, personal phys-
ical, and vehicular forms of anger expression. Pejora-
tive labeling/verbally aggressive, revengeful/retaliatory,
and physically aggressive thinking correlated nega-
tively with adaptive/constructive anger expression. Cop-
ing self-instructions correlated negatively with using
the vehicle to express anger and positively with adap-
tive/constructive expression of anger. The general pattern
was the same for both sexes; however, the correlations
of revengeful/retaliatory and physically aggressive think-
ing with personal physical aggressive driving anger ex-
pression were stronger for men than women. Men also
showed a stronger negative correlation between adap-
tive/constructive anger expression and physically aggres-
sive thinking than women did.
A series of tests for differences in strengths of de-
pendent correlations (Bruning & Kintz, 1997) were con-
ducted on correlations in Table IV to assess whether cer-
tain forms of thinking might be more strongly related
to specific forms of anger expression. Alpha level was
set at p < .05. Such differences, if they were found,
would provide evidence of discriminant validity for the
DAX and DATQ. Pejorative labeling/verbally aggres-
sive and judgmental/disbelieving thinking formed signif-
icantly larger correlations with verbally aggressive anger
expression than with other forms of anger expression,
whereas revengeful/retaliatory formed stronger correla-
tions with using the vehicle as the means of express-
ing anger and coping self-instructions correlated stronger
with adaptive/constructive expression than with other

Table IV. Correlation Between Driving Anger Expression (DAX) and Angry Thoughts (DATQ)

Measure DAX-verbal DAX-physical DAX-vehicle DAX-adapt/cons

DATQ-JD .55(.54) .39(.43) .30(.36) .10(.15)

DATQ-PLVA .75(.77) .55(.59) .55(.53) −.28(−.20)
DATQ-RR .48(.53) .67a (.50)b .81(.80) −.43(−.29)
DATQ-PA .44(.28) .67a (.50)b .56(.42) −.39a (−.09)b
DATQ-CSI −.07(.03) −.10(.06) −.19(−.11) .65(.69)

Note. Correlations for men are outside the parentheses, correlations for women are inside paren-
theses. Different subscripts indicate a significant difference between genders in strengths of cor-
relations ( p < .01). DAX = Driving Anger Expression; Adapt/Cons = Adaptive/Constructive
Expression; DATQ = Driver’s Angry Thoughts Questionnaire; JD = Judgmental/Disbelieving
thinking; PLVA = Pejorative Label/Verbally Aggressive thinking; RR = Revenge and Retalia-
tory thinking; PA = Physically Aggressive thinking; and CSI = Coping self-instruction. r > .18,
p < .01. r > .22, p < .001.
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

94 Deffenbacher, White, and Lynch

Table V. Correlations of Trait Anger and General Anger Expression With sive anger expression with Anger-Control were larger than
Driving Anger Measures the parallel correlations for women. In summary, general
anger and driving anger measures correlated significantly
Measure TAS Anger-in Anger-out Anger-control
and shared, in many cases, considerable common variance.
TAS .17(.31) .74(.64) −.51(−.43) Table VI presents the correlations of trait driving
Anger-in .02(.13) .02(−.07) anger, driving anger expression, angry cognitions, trait
Anger-out −.49(−.57)
anger, and general anger expression with responses from
DAS .50(.41) .19(.18) .39(.28) −.31(−.11)
DAX-VAE .56(.53) .25(.24) .54(.45) −.33(−.22) the Driving Survey. Adaptive/constructive expression of
DAX-PPE .64(.49) .05(.16) .61(.48) −.46a (−.20)b driving anger, controlled expression of anger generally,
DAX-UVE .64(.57) .06(.12) .52(.43) −.43(−.34) and coping-self instructions correlated negatively with ag-
DAX-AC −.36(−.28) .08(.06) −.31(−.27) .47(.52) gressive behavior on the road. All other variables, except
DATQ-JD .35(.33) .20(.20) .36(.26) −.19(.01)
Anger-In with which there was a zero order correlation,
DATQ-PLVA .63(.55) .17(.27) .57(.48) −.31(−.21)
DATQ-RR .66(.54) −.10(.12) .50(.47) −.40(−.34) correlated positively with aggression. The same general
DATQ-PA .65a (.43)b .10(.06) .52a (.27)b −.40(−.19) pattern of correlations was found for risky behavior, albeit
DATQ-CSI −.13(−.09) .16(.02) −.15(−.16) .22(.34) with smaller correlations generally and a few nonsignifi-
cant relationships. Exploration of correlations of variables
Note. Correlations for men are outside the parentheses, correlations for
with crash-related conditions showed that these variables
women are inside parentheses. Different subscripts indicate a signifi-
cant difference between genders in strengths of correlations ( p < .01). were, in general, not highly correlated with anger vari-
DAX = Driving Anger Expression; DATQ = Driver’s Angry Thoughts ables; many correlations were not significant, and ones
Questionnaire; VAE = Verbally Aggressive Expression; PPE = Per- that were significant tended to be small in size. Overall,
sonal Physical Expression; UVE = Use of Vehicle to Express anger; the pattern of correlations was similar for men and women.
JD = Judgmental/Disbelieving thinking; PLVA = Pejorative Label-
However, the correlation of men’s trait anger with aggres-
ing/Verbally Aggressive thinking; RR = Revenge and Retaliatory think-
ing; PA = Physically Aggressive thinking; and CSI = Coping Self- sion was stronger than that correlation for women.
instruction. r > .18, p < .01. r > .22, p < .001.

forms of expression. The pattern for physically aggressive
thinking was less clear. Physically aggressive thinking
formed larger correlations with personal physical and ve-
hicular forms of anger expression than verbal or adap-
tive/constructive forms, but the strengths of relationships
with personal physical and vehicular forms of expression
did not differ.
Table V presents the correlations of trait anger and
general anger expression with trait driving anger, driving
anger expression, and angry cognitions. As found in prior
literature, trait anger correlated positively with Anger-
In and Anger-Out and negatively with Anger-Control.
Anger-In was unrelated to Anger-Out or Anger-Control,
but Anger-Out and Anger-Control correlated negatively.
General anger measures showed many significant cor-
relations with driving anger measures. Trait anger and
Anger-Out were moderately to strongly correlated with
trait driving anger, all forms of expressing anger behind
the wheel, and all angry cognitions, except for coping
self-instructions. Anger-Control showed the reverse pat-
terns of correlations, albeit with somewhat smaller corre-
lations in most cases. Anger-In showed small or nonsignif-
icant correlations with driving anger measures. Strengths
of correlations were similar for men and women, but the
correlations of men’s physically aggressive thinking with
trait anger and Anger-Out and of men’s physically aggres-
Regression Analyses
Correlations showed that forms of driving anger ex-
pression and angry cognitions were meaningfully corre-
lated with each other and with trait driving anger, aggres-
sion and risky behavior on the road, and some crash-related
conditions. Many of these relationships were moderate
to large in size. Moreover, some forms of angry think-
ing were more highly correlated with some forms of ex-
pressing driving anger than others. These findings suggest
that the DAX and DATQ could be useful assessment de-
vices with drivers. However, correlations also showed that
trait driving anger, general trait anger, and general forms
of anger expression were correlated with important vari-
ables such as aggressive and risky behavior, suggesting
that assessment instruments already available might serve
as well as the DAX and DATQ. The value of the DAX and
DATQ would be strengthened if it could be shown that
they added to the prediction of important driving-related
variables above and beyond that provided by other mea-
sures.
To address these issues, a series of hierarchical re-
gressions were conducted. Because some gender differ-
ences were noted in distributions and in strengths of some
correlations, a series of preliminary hierarchical regres-
sions were conducted. Gender was entered on Step 1, the
DAS, TAS, or AX on Step 2, either the DAX or DATQ on
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

Two New Scales for Driving Anger 95

Table VI. Correlations Between Driving Anger Scale, Driver’s Angry Thoughts Questionnaire, Driving Anger Expression, Trait Anger Scale, Anger
Expression Inventory, Aggression, Risky Behavior, and Crash-Related Outcomes

Risky Loss of Loss of Close Moving Minor Major

Measures Aggression behavior concentration control calls violations accidents accidents

DAS .45(.36) .34(.35) .16(.18) .11(.18) .24(.17) .09(.09) .12(.09) .08(.01)

DAX-verbal .49(.56) .29(.31) .20(.13) .16(.06) .15(.17) .01(.09) .07(−.02) .11(.15)
DAX-physical .64(.48) .42(.31) .08(.12) .11(.06) .07(.16) .02(.11) .07(.03) .23(.12)
DAX-vehicle .64(.61) .54(.48) .16(.18) .10(.19) .15(.21) .02(.13) .02(.05) .13(.04)
DAX-adapt/cons −.40(−.40) −.37(−.21) −.19(−.15) −.09(−.12) −.07(−.06) −.01(−.09) .00(.05) .07(−.02)
DATQ-JD .23(.36) .05(.21) .12(.10) .04(.06) .07(.10) −.10(.04) −.04(−.06) .10(.04)
DATQ-PLVA .58(.55) .35(.35) .21(.23) .15(.14) .20(.22) .02(.12) .07(−.01) .10(.10)
DATQ-RR .64(.63) .53(.42) .18(.26) .13(.17) .16(.19) .01(.09) .06(.03) .06(.06)
DATQ-PA .55(.40) .35(.36) .13(.17) .09(.09) .11(.22) −.04(.08) .04(.09) .01(.04)
DATQ-CSI −.20(−.07) −.21(−.01) −.06(.01) −.02(.01) −.03(.09) −.04(−.03) −.05(.00) .06(.05)
TAS .61a (.31)b .46(.42) .15(.15) .11(.16) .19(.23) .03(.15) .11(.13) .11(.08)
Anger-in .02(.11) .06(.18) .18(.16) .20(.19) .20(.25) .08(.07) .14(.07) .00(−.07)
Anger-out .57(.43) .40(.35) .13(.15) .09(.17) .12(.19) −.01(.11) .14(.07) .10(.07)
Anger-control −.36(−.39) −.22(−.25) −.06(−.24) −.03(−.21) −.10(−.19) .08(−.15) −.03(−.10) −.05(−.02)

Note. Correlations for men are outside the parentheses, correlations for women are inside parentheses. Different subscripts indicate a significant
difference between genders in strengths of correlations ( p < .01). DAS = Driving Anger Scale; DATQ = Driver’s Angry Thoughts Questionnaire;
JD = Judgmental/Disbelieving thinking; PLVA = Pejorative Label/Verbally Aggressive thinking; RR = Revenge and Retaliatory thinking; PA =
Physically Aggressive thinking; CSI = Coping Self-instruction; DAX = Driving Anger Expression; Adapt/Cons = Adaptive/Constructive; and TAS =
Trait Anger Scale. r > .18, p < .01. r > .22, p < .001.

Step 3, and All Gender × Step 3 variable interactions on
Step 4. Because none of the interactions were significant
and accounted for at least 1% of the variance, the lower
limit of a small effect size (Cohen, 1988), interactions with
gender were not considered meaningful. A second set of
hierarchical regressions were then run. Gender was en-
tered on Step 1, the DAS, TAS, or AX on Step 2, and the
DAX or DATQ scales were allowed to enter stepwise on
Step 3. Stepwise entry on Step 3 provided a clearer sense
of which DAX or DATQ scales contributed most to added
prediction of the variable of interest. To be retained in the
regression model, DAX or DATQ scales had to be signifi-
cant at p < .01 to control for Type I error and contribute at
least 1% added variance, the lower limit of Cohen’s small
effect size. The latter criterion was designed to limit sta-
tistically significant, but potentially trivial contributions
from being considered as evidence of incremental valid-
ity of the DAX or DATQ. Finally, since correlations in
Table VI showed that crash-related conditions were not
well predicted by any variable, many regression models
for these variables were highly similar and redundant. To
reduce redundancy, contributions to regressions on crash-
related conditions will be summarized in the text, whereas
Table VII summarizes regression models for aggressive
and risky behavior.
To assess whether driving anger expression (DAX
scales) added to the prediction of aggressive and risky be-
havior beyond that provided by level of driving anger or
general anger, hierarchical regression analyses entered ei-
ther the DAS or the TAS on Step 2 and DAX scales on
Step 3 (Table VII). Gender did not contribute to regres-
sions on aggression; however, the DAS and TAS contribute
large amounts of variance in aggression. Three DAX
scales, however, contributed an additional 33.5 and 17.6%
additional variance, respectively. Use of the Vehicle to Ex-
press Anger contributed large and moderate amounts of
variance, Personal Physical Expression contributed mod-
erate and small amounts, and Adaptive/Constructive Ex-
pression an additional small amount. Risky behavior fol-
lowed a somewhat similar pattern. Gender contributed a
small amount of variance, whereas the DAS and TAS con-
tributed moderate and large amounts of variance. DAX
scales, however, contributed an additional 17.8 and 9.7%
of explained variance. Use of the Vehicle to Express Anger
was the primary significant contributor to both equations.
The added value and incremental validity of the DAX
would also be enhanced if DAX scales contributed mean-
ingful variance above and beyond that provided by a mea-
sure of general anger expression (the AX). To address this
issue, another set of hierarchical regressions were con-
ducted in which gender was entered on Step 1, AX scales
on Step 2, and DAX scales stepwise on Step 3. General
anger expression (AX) contributed large amounts of vari-
ance to the prediction of aggressive and risky behavior,
but DAX scales contributed additional variance beyond
the AX. Use of the Vehicle to Express Anger contributed
additional large amounts of variance to the prediction
of aggressive (17.2%) and risky behavior (14.0%), and
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004 10:51 Style file version June 25th, 2002

96 Deffenbacher, White, and Lynch

Table VII. Hierarchical Regressions With Gender on Step 1; DAS, TAS, or AX on Step 2; and DAX or DATQ Stepwise on Step 3

Measure Variables in the equation R2

Equations with DAX on Step 3

Aggressive behavior Gender (β = −.12, 1.3%), DAS (β = .41, 16.3%∗∗∗ ), UVE (β = .56, .511∗∗∗
24.7%∗∗∗ ), PPE (β = .31, 5.8%∗∗∗ ), and AC (β = −.19, 3.0%∗∗∗ )
Risky behavior Gender (β = −.16, 2.6%∗∗∗ ), DAS (β = .33, 11.0%∗∗∗ ), and UVE (β =.46, 16.8%∗∗∗ ) .304∗∗∗
Aggressive behavior Gender (β = −.12, 1.3%), TAS (β =.58, 32.5%∗∗∗ ), UVE (β =.44, .515∗∗∗
11.8%∗∗∗ ), PPE (β = .26, 3.6%∗∗∗ ), and AC (β = −.17, 2.2%∗∗∗ )
Risky behavior Gender (β = −.16, 2.6∗∗∗ ), TAS (β =.44, 18.8%∗∗∗ ), and UVE (β =.40, 9.7%∗∗∗ ) .311∗∗∗
Aggressive behavior Gender (β = −.12, 1.3%), AX (27.0%∗∗∗ ), UVE (β = .49, 17.2%∗∗∗ ) .509∗∗∗
PPE (β =.26, 3.6%∗∗∗ ), and AC (β = −.17, 1.8%∗∗ )
Risky behavior Gender (β = −.16, 2.6%∗∗∗ ), AX (14.8%∗∗∗ ), and UVE (β = .44, 14.0%∗∗∗ ) .312∗∗∗
Equations with DATQ on Step 3
Aggressive behavior Gender (β = −.12, 1.3%), DAS (β =.41, 16.3%∗∗∗ ), RR (β =.58, .484∗∗∗
25.9%∗∗∗ ), PLVA (β = .23, 3.0%∗∗∗ ), and JD (β = −.18, 1.8%∗∗ )
Risky behavior Gender (β = −.16, 2.6%∗∗∗ ), DAS (β = .33, 11.0%∗∗∗ ), and RR (β = .44, 14.5%∗∗∗ ) .291∗∗∗
Aggressive behavior Gender (β = −.12, 1.3%), TAS (β =.58, 32.5%∗∗∗ ), RR (β = .47, .533∗∗∗
12.7%∗∗∗ ), and PLVA (β = .17, 2.0%∗∗ )
Risky behavior Gender (β = −.16, 2.6∗∗∗ ), TAS (β = .36, 18.8%∗∗∗ ), and RR (β =.36, 7.7%∗∗∗ ) .470∗∗∗

Note. The first number within the parentheses is the standardized β weight for the variable entered on that step. The second number within the parentheses
is the percent added variance or 1R 2 provided by the variable when entered on that step. Asterisks within parentheses reflect the significance of the F
to enter; asterisks after the R 2 reflect the significance of the total variance accounted for by the equation. DAS = Driving Anger Scale; TAS = Trait
Anger Scale; DAX = Driving Anger Expression Inventory; UVE = Use of the Vehicle to Express anger; PPE = Personal Physical Expression of anger;
AC = Adaptive/Constructive Expression of anger; DATQ = Driver’s Angry Thoughts Questionnaire; RR = Revenge and Retaliatory thinking; PLVA
= Pejorative Labeling/Verbally Aggressive thinking; and JD = Judgmental and Disbelieving thinking.
∗∗ p < .01. ∗∗∗ p < .001.

Personal Physical Expression and Adaptive/Constructive
Expression contributed additional small amounts to the
prediction of aggression (3.6 and 1.8%) for a total of
22.6% additional predicted variance of aggression for the
DAX.
The validity and utility of the DATQ would be
strengthened further if this cognitive measure added to
the prediction of driving-related variables above and be-
yond that provided by measures of anger such as the DAS
and TAS. As in earlier analyses, hierarchical regressions
(Table VII) were conducted with gender entered on Step 1,
the DAS or TAS on Step 2, and DATQ scales stepwise on
Step 3. Gender and anger measures accounted for variance
as in earlier regression models, but revengeful/retaliatory
thinking accounted for large amounts of additional vari-
ance in aggressive and risky behavior on the road (25.9 and
12.7% to aggression and 14.5 and 7.7% to risky behavior,
respectively). Pejorative labeling and verbally aggressive
thinking also added variance for aggressive behavior (3.0
and 2.0%).
Regarding regressions on crash-related conditions,
gender contributed significantly only to the regression on
moving violations (2.2%), where it was the only signifi-
cant predictor. Men averaged more moving violations than
women. Regressions involving the DAS and TAS (1–5%
additional variance) and the AX scales together (2–8%
additional variance) showed that they contributed small to
moderate amounts of variance to equations for losses of
concentration, minor losses of vehicular control, and close
calls, and minor accidents, but did not contribute to the
prediction of moving violations or major accidents. DAX
scales also contributed to some of these regressions. Adap-
tive/Constructive Expression contributed an additional 1–
2% to the prediction of losses of concentration, and Per-
sonal Physical Expression was the sole significant predic-
tor of major accidents (3–3.5% of variance). Pejorative
Labeling/Verbally Aggressive thinking contributed an ad-
ditional 1–3% predicted variance for losses of concentra-
tion and close calls.
DISCUSSION
Limitations of the Study
This study has two major limitations. First, it em-
ployed a college sample. The degree of generalizability to
older, younger, more diverse drivers, or specialized pop-
ulations such as court-referred drivers is unknown. How-
ever, college students are a meaningful group in their own
right. They represent large group of young adults who, as
a group, experience considerable anger while driving and
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578
Two New Scales for Driving Anger
for whom auto accidents are a leading source of injury
and death. Thus, they are a meaningful group upon which
to develop and validate assessment strategies that could
be of value in other populations. Second, the study re-
lied entirely on self-report measures. Self-report is most
appropriate for phenomena such as angry feelings and
cognitions, because these are internal conditions, often
best reported by the individual. However, other parameters
could be bolstered by the use of behavioral observation,
reports of others, and archival records (e.g., records of
traffic citations and accidents). However, within the limits
of self-report measures, results provided validation of two
instruments that could enhance other information about
angry drivers.
Gender Differences
This study found a greater number of gender differ-
ences than prior studies. For example, men reported more
personal physical and vehicular anger expression and
less adaptive/constructive anger expression than women,
whereas in a prior study (Deffenbacher, Lynch, et al.,
2002) men and women differed only on personal physical
expression. Gender differences on revengeful/retaliatory
and physically aggressive thinking replicated the earlier
study (Deffenbacher et al., in press), and gender was also
the sole predictor of moving violations, with men receiv-
ing more tickets than women. Additionally, some gen-
der differences were noted in the strength of correlations,
where such differences were not found in prior studies
(Deffenbacher et al., in press; Deffenbacher, Lynch, et al.,
2002). Although there were many nonsignificant correla-
tions and occasional inconsistencies, correlations for men
and women generally followed the same pattern, but a
small number of correlations between aggressive anger
expression, aggressive thinking, and aggression tended to
be larger for men than women. However, regression anal-
yses in which interactions of variables with gender were
included showed that these differences did not influence
major findings.
Correlational Findings: Relationships Among Scales
for Each Measure and Convergent Validity
Relationships among scales on the DAX repli-
cated those found previously (Deffenbacher et al., 2001,
Deffenbacher, Lynch, et al., 2002). Specifically, verbal,
physical, and vehicular forms of anger expression corre-
lated positively with each other and minimally or nega-
tively with adaptive/constructive expression. The degree
and pattern of correlations suggested three things. First,
January 28, 2004 10:51 Style file version June 25th, 2002
97
measures were independent enough to suggest that they
assess different, but correlated elements of driving anger
expression. Second, aggression and positive coping with
anger are not the ends of a continuum. Although they
were not completely unrelated, the zero order or small
degree of correlation between adaptive/constructive and
more aggressive forms of expression suggests that they
are somewhat different constructs. Clinically, this suggests
that the aggressive and constructive expression should be
assessed and targeted separately and that it cannot be as-
sumed that the reduction of aggressive forms of expression
will automatically lead to positive means of dealing with
anger and frustration on the road. Third, the relatively low
degree of correlation of adaptive/constructive expression
with aggression and risky behavior suggests that safe driv-
ing practices, both generally and when angry, should also
be assessed and correlated to see if adaptive/constructive
forms of handling anger correlate with safe driving.
The relationship among scales on the DATQ also
replicated those found earlier (Deffenbacher et al., in
press). Judgmental/disbelieving thinking correlated posi-
tively with all forms of angry thinking. The three most ag-
gressive forms of angry cognitions (revengeful/retaliatory,
physically aggressive, and pejorative labeling/verbally ag-
gressive) correlated positively with each other, but once
again the degree of correlation suggested that they were as-
sessing related, but separable dimensions of angry thought
processes. Coping self-instructional thinking correlated
minimally with or formed small negative correlations with
the three more aggressive patterns of thinking. Much
like the correlation of adaptive/constructive expression
with aggressive forms of anger expression, this pattern
of correlation suggested that positive, adaptive cognitions
are somewhat orthogonal to more hostile and aggressive
thinking and that changing one will not necessarily lead
to changes in the other. Both types of thinking may need
separate attention in assessment and treatment.
Correlations of DAX and DATQ scales with other
variables replicated earlier findings (Deffenbacher et al.,
2001; Deffenbacher, Lynch, et al., 2002) and provided
evidence of convergent validity. Specifically, aggressive
forms of anger expression (verbal, physical, and vehic-
ular) and angry cognitions (pejorative labeling/verbally
aggressive, revengeful/retaliatory, and physically aggres-
sive) correlated positively with each other, trait driving
anger, aggressive and risky behavior on the road, gen-
eral trait anger, and aggressive general anger expression
(Anger-Out). These measures also tended to correlate neg-
atively with adaptive/constructive driving anger expres-
sion, coping self-instruction, and controlled anger expres-
sion generally (Anger-Control), whereas the latter three
measures tended to correlate positively with each other.
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578 January 28, 2004
98
The overall pattern of correlations supported convergent
validity for the DAX and DATQ, as they correlated logi-
cally and meaningfully with other measures.
Added Value: Evidence of Discriminant and
Incremental Validity
The three most aggressive types of angry cogni-
tions (pejorative labeling/verbally aggressive, revenge-
ful/retaliatory, and physically aggressive) correlated
strongly with driving anger and aggressive anger ex-
pression. This suggests that these types of thoughts are
associated with anger arousal when driving and may chan-
nel behavior toward aggression. However, it was the pat-
tern of differential strengths of correlations of forms of
thinking with different types of anger expression that
suggests an even stronger link. Specifically, pejorative
labeling/verbally aggressive thinking was most strongly
associated with verbally aggressive expression, revenge-
ful/retaliatory thinking most strongly associated with us-
ing the vehicle to express anger, and revengeful/retaliatory
and physically aggressive thinking more with physically
aggressive expression. On the more positive side, coping
self-instructional thinking and adaptive/constructive ex-
pression were also most strongly linked. Somewhat sim-
ilar item content on the DATQ and DAX may account
for some of this association, but the patterns of corre-
lations suggest that specific types of angry thinking are
linked differentially with different forms of anger expres-
sion, which, in turn, may focus and direct angry feelings
and aggressive urges toward specific forms of expression.
These patterns of differential strengths of correlations sup-
port the discriminant validity of the DATQ and DAX and
suggest that specific thought-action sequences should be
assessed and treated together.
Regression analyses supported the incremental va-
lidity of the DAX and DATQ. Forms of anger expression,
especially using the vehicle to express anger, personal
physical aggressive expression, and adaptive/constructive
expression, consistently added explained variance to re-
gressions on aggressive and risky behavior, above and be-
yond that contributed by trait driving anger (DAS), general
trait anger (TAS), and general forms of anger expression
(AX). Angry thinking also contributed to the prediction
of risky and aggressive behavior above and beyond that
provided by the DAS and TAS, with revengeful/retaliatory
thinking and pejorative labeling/verbally aggressive think-
ing making the largest contributions. These findings sug-
gest that the DAX and DATQ added additional, predic-
tive information not readily provided by other established
measures, which supports their value as an assessment
instruments for driving anger.
10:51 Style file version June 25th, 2002
Deffenbacher, White, and Lynch
In summary, both the DAX and DATQ hold promise
for researchers and clinicians working with driving anger.
They assess important elements of an angry driver’s expe-
rience, namely what he/she may think and how he/she may
express anger. The strength and pattern of correlations sup-
port their validity in relation to anger, aggression, and risky
behavior on the road and, in some cases, to crash-related
events. Evidence of discriminant and incremental validity
further strengthens their value as instruments for assessing
angry cognitions and forms of expression of anger while
driving.
ACKNOWLEDGMENTS
This study was supported, in part, by Grant
R49/CCR811509 from the Centers for Disease Control
and Prevention and by 5 P50 DA07074-10 from the Na-
tional Institute on Drug Abuse. Its contents are solely the
responsibility of the authors and do not necessarily repre-
sent the official views of the Centers for Disease Control
and Prevention.
REFERENCES
American Automobile Association. (1997). Aggressive driving: Three
studies. Washington, DC: American Automobile Association Foun-
dation for Traffic Safety.
Bruning, J. L., & Kintz, B. L. (1977). Computational handbook of statis-
tics. Glenview, IL: Scott, Foresman and Company.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
Deffenbacher, J. L. (2000). The Driving Anger Scale (DAS). In J.
Maltyby, C. A. Lewis, & A. Hill (Eds.), Comissioned reviews of 250
psychological tests (pp. 287–292). Lampeter, UK: Edwin Mellen.
Deffenbacher, J. L., Deffenbacher, D. M., Lynch, R. S., & Richards,
T. L. (2003). Anger, aggression, and risky behavior: A comparison
of high and low anger drivers. Behaviour Research and Therapy,
41, 701–718.
Deffenbacher, J. L., Filetti, L. B., Lynch, R. S., Dahlen, E. R., & Oetting,
E. R. (2002). Cognitive-behavioral treatment of high anger drivers.
Behaviour Research and Therapy, 40, 717–737.
Deffenbacher, J. L., Filetti, L. B., Richards, T. L., Lynch, R. S., & Oetting,
E. R. (2003). Characteristics of two groups of angry drivers. Journal
of Counseling Psychology, 50, 123–132.
Deffenbacher, J. L., Huff, M. E., Lynch, R. S., Oetting, E. R., & Salvatore,
N. F. (2000). Characteristics and treatment of high anger drivers.
Journal of Counseling Psychology, 47, 5–17.
Deffenbacher, J. L., Lynch, R. S., Deffenbacher, D. M., & Oetting, E. R.
(2001). Further evidence of reliability and validity for the Driving
Anger Expression Inventory. Psychological Reports, 89, 535–540.
Deffenbacher, J. L., Lynch, R. S., Filetti, L. B., Dahlen, E. R., & Oetting,
E. R. (2003). Anger, aggression, risky behavior, and crash-related
outcomes in three groups of drivers. Behaviour Research and Ther-
apy, 41, 333–349.
Deffenbacher, J. L., Lynch, R. S., Oetting, E. R., & Swaim, R. C. (2002).
The Driving Anger Expression Inventory: A measure of how people
express their anger on the road. Behaviour Research and Therapy,
40, 717–737.
P1: GAD
Journal of Psychopathology and Behavioral Assessment (JOBA) pp1116-joba-480578
Two New Scales for Driving Anger
Deffenbacher, J. L., Oetting, E. R., & Lynch, R. S. (1994). Devel-
opment of a driving anger scale. Psychological Reports, 74, 83–
91.
Deffenbacher, J. L., Oetting, E. R., Lynch, R. S., & Morris, C. D. (1996).
The expression of anger and its consequences. Behaviour Research
and Therapy, 34, 575–590.
Deffenbacher, J. L., Oetting, E. R., Thwaites, G. A., Lynch, R. S., Baker,
D. A., Stark, P. S., et al., (1996). State-trait anger theory and the
utility of the Trait Anger Scale. Journal of Counseling Psychology,
43, 131–148.
Deffenbacher, J. L., Petrilli, R. T., Lynch, R. S., Oetting, E. R., & Swaim,
R. C. (2003). The Driver’s Angry Thoughts Questionnaire: A mea-
sure of angry cognitions when driving. Cognitive Therapy and Re-
search, 27, 383–402.
Galovski, T. E., & Blanchard, E. B. (2002). The effectiveness of a
brief psychological intervention on court-referred and self-referred
aggressive drivers. Behaviour Research and Therapy, 40, 1385–
1402.
Galovski, T., Blanchard, E. B., & Veazey, C. (2002). Intermittent ex-
plosive disorder and other psychiatric comorbidity among court-
referred and self-referred aggressive drivers. Behaviour Research
and Therapy, 40, 641–651.
Hemenway, D., & Solnick, S. (1993). Fuzzy dice, dream cars and inde-
cent gestures: Correlates of driving behaviour? Accident Analysis
and Prevention, 25, 161–170.
Jacobs, G. A., Latham, L. E., & Brown, M. S. (1988). Test–retest reli-
ability of the State-Trait Personality Inventory and the Anger Ex-
pression Scale. Anxiety Research, 1, 363–365.
January 28, 2004 10:51 Style file version June 25th, 2002
99
Lajunen, T., & Parker, D. (2001). Are aggressive people aggressive
drivers? A study of the relationship between self-reported general
aggressiveness, driver anger and aggressive driving. Accident Anal-
ysis and Prevention, 33, 243–255.
Lajunen, T., Parker, D., & Stradling, S. G. (1998). Dimensions of driver
anger, aggressive and highway code violations and their mediation
by safety orientation in UK drivers. Transportation Research, Part
F 1,107–121.
Martinez, R. (1997, July). Statement of the Honorable Recardo Martinez,
M. D., of the National Highway Traffic Safety Administration. Sub-
committee on Surface Transportation, Committee on Transporta-
tion and Infrastructure, U.S. House of Representatives, Washington,
D.C.
Morris, C. D., Deffenbacher, J. L., Lynch, R. S., & Oetting, E. R. (1996,
August). Anger expression and its consequences. Paper presented
at the 104th Annual Convention of the American Psychological
Association, Toronto, Ontario, Canada.
Snyder, D. S. (1997, July). Statement of David S. Snyder, Assistant
General Counsel of the American Insurance Association. Sub-
committee on Surface Transportation, Committee on Transporta-
tion and Infrastructure, U.S. House of Representatives, Washington,
D.C.
Spielberger, C. D. (1988). State-Trait Anger Expression Inventory (1st
ed.). Odessa, FL:Psychological Assessment Resources.
Spielberger, C. D. (1999). State-Trait Anger Expression Inventory (2nd
ed.). Odessa, FL:Psychological Assessment Resources.
Underwood, G., Chapman, P., Wright, S., & Crundall, D. (1999). Anger
while driving. Transportation Research, Part F 2, 55–68.